Analysis report
Employee Selection Ratings
Data from mldata.org, TU Berlin, published under https://diafunc.com/licences/view-only.
- Target
- Target
- Analyzed
- 2026-10-28 14:02 UTC
Summary
The model analyzes Target using three inputs and finds that the strongest input is In4, which makes up 71% of the model's reliance. The model explains 85% of the variation in Target, and moving In4 from 5.00 to 6.00 raises the model's output by about 0.45 on average, with everything else as observed. Compared with a Poisson regression on the numeric inputs, the model removes 16 fewer points of Poisson deviance and the two are about level, with the model ahead on 49% of the held-out rows.
The model did not beat the baseline Poisson regression, so its findings are not an advance. The check rests on 98 held-out rows and 3.1% lay outside the range the model was trained on. The held-out split is likely contaminated by the 65% exact duplicate rows in the data. Validate on a fresh sample before relying on the model for decisions.
Formula
Target = -1.6 + 0.37 × In2 + 0.44 × In3 + 0.45 × In4Held-out diagnostics
| Target | Quality | Error | R² | MAE | Poisson deviance | RMSE | Dispersion | Generalization | Rows |
|---|---|---|---|---|---|---|---|---|---|
| Target | 83.92% | 15.1% | 0.8483 | 0.4284 | 0.06662 | 0.5824 | 0.06647 | generalizing | 98 |
- Target · Quality 95% CI 78% – 88.38%
- Target · Error 95% CI 10.91% – 20.66%
- Target · R² 95% CI 0.79 – 0.8887
- Target · MAE 95% CI 0.355 – 0.5035
- Target · Poisson deviance 95% CI 0.0466 – 0.08695
- Target · RMSE 95% CI 0.4774 – 0.6739
- Target · Dispersion 95% CI 0.04709 – 0.08675
Concerns
CRITICAL data
Data has 65% exact duplicate rows: the held-out split is likely contaminated
317 of 488 rows (65%) are exact duplicates of at least one other row (largest duplicate group: 11 identical rows). At this density some duplicate pairs very likely straddle the held-out split, in which case the held-out R² or accuracy overstates how the model does on rows it has not seen. Deduplicate before deploying: drop the duplicates outright, or, where they encode intentional row weights, aggregate them into one weighted row. An analysis of the deduplicated data reads the headline metric without the inflation.
WARNING model
The model does not beat a Poisson regression on the numeric inputs on `Target`
On the held-out rows the model leaves 16% more Poisson deviance than a Poisson regression on the numeric inputs. The interval of the margin includes zero: on this many held-out rows the two are not told apart. The formula adds little over the reference predictor; whether it is worth its complexity depends on the use. A plain regression on the same inputs does as well or better, so the inputs do carry signal; the search did not find a better formula in the time it had. Row by row the two are about level.
Analysis
The report opens with the Headline and the Concerns, followed by the Discovered Relations when the analysis has targets and by Questions & Answers when questions were asked. Three sections carry the detail: Model presents one model per target, Implications states where the models hold and what to weigh before relying on them, and Data describes the table as supplied, column by column, for reference.
Generated 2026-10-28 14:02 UTC, analysis 764da0d0-48b7-11f1-8087-d378640fe787, 1 target, time budget 1 h 0 min, compute budget 500 credits, 488 rows. The Provenance & Audit chapter holds the full record.
Validity
Validity: valid. Nothing in the data or the formula speaks against relying on this result.
How to read this
Whether this result is one to rely on. Quality says how closely the formula follows the data; validity says whether that closeness means anything. A formula that only restates its target, or that reads a column which already holds the answer, can reach a high quality and still say nothing new about the data.
What it found
Target starts from a base of -1.6 and rises with In4, In3 and In2.
- In4 accounts for 71% of what the formula explains. In4 accounts for 71% of what the formula explains and In3 for 21%. In2 adds 7.9%.
- The model's output rises with In4, In3 and In2 at steady rates. One more unit of In4 goes with the model's output higher by 0.45; of In3 by 0.44 and of In2 by 0.37. The formula adds these up, so no rate depends on the values of the other inputs.
- In1 is not in the formula. It has the second-strongest correlation with Target in the table (0.75). Rows with more of it have more In2 (correlation 0.76), and In2 is in the formula.
- The formula does worse than a Poisson regression. The formula removes 85% of the Poisson deviance of always guessing the average count. Its Poisson deviance is 16% higher than a Poisson regression's. That was measured on 98 rows it never saw. The formula typically misses Target by 0.428.
Summary
What to do next: Is not told apart from the best yardstick by 16% of its Poisson deviance.
6 chapters, each headed by what it found.
S.1 In4 carries most of the formula
What the formula relies on. Shares of the formula: In4 71%, In3 21%, In2 7.9%.
The model's output rises with In4 on every one of the 1,170 steps between observed values. The model's output rises with In3 on every one of the 1,170 steps between observed values. The model's output rises with In2 on every one of the 1,560 steps between observed values. Moving In4 from 5.00 to 6.00, its first and third quartile, with every other input as observed raises the model's output by 0.45 on average. These are measurements on the rows the search trained on, inside the observed ranges; outside them nothing is claimed.
| Input | Share of importance | Permutation importance | Sensitivity | Residual correlation | Occurrences | Model's output as the input rises | Change, first to third quartile |
|---|---|---|---|---|---|---|---|
| In4 | 70.85% | 0.7953 | 0.45 | -0.2965 | 1 | Rises (1170 of 1170 steps) | +0.45 (5 to 6) |
| In3 | 21.22% | 0.2382 | 0.44 | 0.06217 | 1 | Rises (1170 of 1170 steps) | +0.44 (5 to 6) |
| In2 | 7.93% | 0.0890 | 0.37 | 0.3862 | 1 | Rises (1560 of 1560 steps) | +0.74 (4 to 6) |
- Rows the search trained on, measured
- 195
- Inputs the formula does not read
- In1
Read as a sum, the formula has 4 additive terms: 3 that read an input and a constant base. All that read an input are needed to reach 100% of the variation of the model's output on the 195 rows the search trained on: the largest are 0.37 × In2 (39%), 0.44 × In3 (36%) and 0.45 × In4 (25%). A share describes the model's output, not the target. Each curve below shows one term's response to its free inputs, with the other inputs held at the training median.
- Additive
- Yes
- Rows the search trained on, measured
- 195
- Terms in the formula
- 4
- Notation
- -1.6 + 0.37 × In2 + 0.44 × In3 + 0.45 × In4
- Target
- Target
S.2 A Poisson regression does better here
How well it holds. The formula removes 85% of the Poisson deviance of always guessing the average count. Its Poisson deviance is 16% higher than a Poisson regression's.
How this was checked
- Measured on 98 rows the formula never saw (20% of 488); the other 390 built it.
- Compared on the same rows against always guessing the average count and a Poisson regression.
- Intervals from 1,000 bootstrap resamples of those rows.
On the held-out rows the model explains 85% of the variation in the counts (R² 0.848); its mean absolute error is 0.428. Mean Poisson deviance 0.0666, with under-dispersion (Pearson ratio 0.0665 < 1). The counts vary less than Poisson expects, often a bounded or capped process. RMSE (root mean squared error) on counts is 0.582.
- Dispersion
- 0.06647 (95% CI 0.04709 – 0.08675)
- Error
- 15.1% (95% CI 10.91% – 20.66%)
- MAE
- 0.4284 (95% CI 0.355 – 0.5035)
- N
- 98
- Training rows
- 390
- Poisson deviance
- 0.06662 (95% CI 0.0466 – 0.08695)
- Quality
- 83.92% (95% CI 78% – 88.38%)
- R²
- 0.8483 (95% CI 0.79 – 0.8887)
- RMSE
- 0.5824 (95% CI 0.4774 – 0.6739)
- Target
- Target
The model does not beat the simple yardsticks: a Poisson regression on the numeric inputs is ahead by 16% of its Poisson deviance on the held-out rows. The model's complexity is not buying anything yet. A plain regression on the same inputs does as well or better, so the inputs do carry signal; the search did not find a better formula in the time it had. Of the 98 held-out rows where the two differ, the model is closer on 49% (95% interval 39% to 59%). The interval of the margin includes zero: on this many held-out rows the two are not told apart.
- Best yardstick's score
- 0.05761 (95% CI 0.03995 – 0.07746)
- Effect size d
- -0.0889 (95% CI -0.2698 – 0.1074)
- Effect size d magnitude
- negligible
- Margin over the best yardstick
- -0.1564 (95% CI -0.4778 – 0.0851)
- Margin's interval stays above zero
- No
- Model
- 0.06662 (95% CI 0.0466 – 0.08695)
- Held-out rows
- 98
- Held-out rows without an answer
- 0
- Target
- Target
- Rows where the model is closer
- 48.98% (95% CI 39.31% – 58.73%)
- Rows compared
- 98
- Rows level
- 0
- Regression yardstick trained with gaps filled (too few complete rows)
- No
- Training rows of the regression yardstick
- 390
The model explains held-out rows about as well as training rows (held-out R² 0.848, gap between training and held-out R² -0.0152): it generalizes. The structure the formula encodes carries from the training rows to the held-out rows.
- Generalization gap
- -0.01517
- N test
- 98
- Target
- Target
- Test error
- 0.4284 (95% CI 0.355 – 0.5035)
- Test R²
- 0.8483 (95% CI 0.79 – 0.8887)
- Train error
- 0.1493
- Train R²
- 0.8331
- Verdict
- Generalizing
S.3 No range of its inputs where it is markedly weaker
Where it is weak. The largest miss on a held-out row is 1.78 (recorded 9, predicted 7.22). 3.1% of the held-out rows are unlike the rows it learned from.
Segment in2 in the third quarter (5) performs in line with the aggregate (20 held-out rows): segment R² 0.667 against aggregate R² 0.848 (gap -0.181). No quality drift detected.
| Segment | Rows | R² | Gap to all rows | Verdict |
|---|---|---|---|---|
| In2 = q3 | 20 | 0.6671 | -0.1812 | In line with all rows |
| In3 = q4 | 47 | 0.6355 | -0.2128 | In line with all rows |
| In1 = q4 | 59 | 0.7204 | -0.1279 | In line with all rows |
| In4 = q4 | 52 | 0.6593 | -0.1889 | In line with all rows |
| In3 = q1 | 25 | 0.4853 | -0.3630 | In line with all rows |
| In3 = q3 | 26 | 0.5228 | -0.3255 | In line with all rows |
| In2 = q4 | 49 | 0.7392 | -0.1091 | In line with all rows |
| In4 = q2 | 33 | 0.7665 | -0.0817 | In line with all rows |
| In1 = q1 | 25 | 0.7738 | -0.0744 | In line with all rows |
| In2 = q2 | 14 | 0.6528 | -0.1955 | Too few rows to judge |
The first 10 rows of 13. The other 3 are in the notebook.
- Aggregate MAE
- 0.4284
- Aggregate R²
- 0.8483
- Measure
- r2
- Target
- Target
The worst held-out row was missed by 1.78 (target's scale). The 10 rows with the largest absolute residuals out of 98 held-out rows are shown. Each card breaks the failure down into per-term contributions and input values, so the failure mode is inspectable. A large residual with one concentrated term contribution means one term got this row wrong; a large residual with balanced contributions means a structural mismatch.
| Actual | Predicted | Residual | In2 | In3 | In4 |
|---|---|---|---|---|---|
| 9.000 | 7.220 | +1.780 | 7 | 7 | 7 |
| 5.000 | 6.770 | -1.770 | 7 | 7 | 6 |
| 9.000 | 7.300 | +1.700 | 6 | 7 | 8 |
| 7.000 | 5.600 | +1.400 | 5 | 5 | 7 |
| 3.000 | 4.260 | -1.260 | 5 | 4 | 5 |
| 4.000 | 5.220 | -1.220 | 4 | 6 | 6 |
| 5.000 | 3.890 | +1.110 | 4 | 4 | 5 |
| 4.000 | 5.060 | -1.060 | 6 | 6 | 4 |
| 4.000 | 4.990 | -0.990 | 7 | 5 | 4 |
| 2.000 | 2.990 | -0.990 | 4 | 4 | 3 |
- Additive
- Yes
- N extrapolated
- 0
- Held-out rows
- 98
- Held-out rows for the interval
- 98
- Residual P5
- -0.99
- Residual P95
- 0.926
- Residual std
- 0.5837
3 of 98 held-out rows (3.1%) lie further from the training rows than 99 in 100 rows like them would, measured on the 3 inputs the model reads (in2, in3, in4). That is a small minority; the held-out figures describe the model on rows like the ones it was trained on.
| Target | Inputs measured | Line | Held-out rows | Beyond the line | Share |
|---|---|---|---|---|---|
| Target | In2, In3, In4 | 3.368 | 98 | 3 | 3.06% |
| Count | From | To |
|---|---|---|
| 0 | 0 | 0.135 |
| 0 | 0.135 | 0.269 |
| 0 | 0.269 | 0.404 |
| 4 | 0.404 | 0.539 |
| 0 | 0.539 | 0.674 |
| 8 | 0.674 | 0.808 |
| 13 | 0.808 | 0.943 |
| 13 | 0.943 | 1.078 |
| 0 | 1.078 | 1.213 |
| 0 | 1.213 | 1.347 |
The first 10 rows of 24. The other 14 are in the notebook.
S.4 Not ready to use yet: 1 critical concern
What to watch. First: Data has 65% exact duplicate rows: the held-out split is likely contaminated.
| Severity | Section | Concern | Detail |
|---|---|---|---|
| Critical | Data | Data has 65% exact duplicate rows: the held-out split is likely contaminated | 317 of 488 rows (65%) are exact duplicates of at least one other row (largest duplicate group: 11 identical rows). At this density some duplicate pairs very likely straddle the held-out split, in which case the held-out R² or accuracy overstates how the model does on rows it has not seen. Deduplicate before deploying: drop the duplicates outright, or, where they encode intentional row weights, aggregate them into one weighted row. An analysis of the deduplicated data reads the headline metric without the inflation. |
| Warning | Model | The model does not beat a Poisson regression on the numeric inputs on Target | On the held-out rows the model leaves 16% more Poisson deviance than a Poisson regression on the numeric inputs. The interval of the margin includes zero: on this many held-out rows the two are not told apart. The formula adds little over the reference predictor; whether it is worth its complexity depends on the use. A plain regression on the same inputs does as well or better, so the inputs do carry signal; the search did not find a better formula in the time it had. Row by row the two are about level. |
The analysis ran with the configuration below. The checks and the verdict of this chapter are read against it.
- Targets: 1 target:
target - Rows: 488 rows
- Per-Segment Predictions: not produced, because no segment column was chosen
- Fairness: not assessed, because the data has no categorical column other than the target
- Out-of-Distribution: held-out rows are scored against the training rows; predictions made outside this report are not scored
- Provenance & Audit: recorded in the chapter of that name
Across 4 checks (3 pass, 1 warns, 0 fail, 1 critical concern), every check passes or warns, but a critical concern is open: Not ready to use yet. The Concerns chapter names what is open; it needs addressing before the predictions are used.
| Name | Status | Detail |
|---|---|---|
| Generalization | Pass | held-out quality holds up against training quality |
| Baseline Lift for Target | Warn | is not told apart from the best yardstick by 16% of its Poisson deviance |
| Model Constraints | Pass | every held-out prediction is finite and within the training range |
| Out-of-Distribution | Pass | 3.1% of held-out rows are out of distribution |
- N critical concerns
- 1
- N fail
- 0
- N pass
- 3
- N warn
- 1
- Overall verdict
- Not ready
S.5 A larger held-out set would firm up the figures
What to collect next. The figures rest on 98 held-out rows.
- Collect new, non-duplicate rows for Target. Reason: the data has 65% exact duplicate rows, so the held-out split is likely contaminated, which limits what the model's quality figure says about generalization.
- Inspect why the model does not beat a Poisson regression on the numeric inputs for Target. Reason: the model does not beat a Poisson regression on the numeric inputs on Target, and the lift of -16 points (Poisson deviance removed) is negligible.
S.6 It shows how Target moves with its inputs in these rows, not what causes it
What this cannot tell you. It was built on In4 from 2 to 8, In3 from 2 to 8 and In2 from 0 to 9; outside those ranges it has not been checked. 4 slices of the data had too few held-out rows to judge.
Diafunc found a formula that captures the structure in the training data. That structure is correlative: it states what predicts what in this data. It does not establish that any input causes the target.
A correlative model is the right tool for predicting the target on new rows like these, for exploring which inputs carry information about the target, and for an early-stage diagnostic or screening tool.
A correlative model is not the right tool for claiming that intervening on an input will change the target, for attributing a target's values to specific causes for accountability, or for choosing between policies that act on the inputs.
The Quality by Input Range, Fairness and Concerns chapters show the most common confounding patterns in the data.
The model is correlative: it captures which columns predict the target in the training data, not whether changing one would change the other. Predictions are reliable for inputs that look like the training distribution; they are not evidence that intervening on an input will change the target. The 4 upgrade pathways below show what evidence would support a causal claim.
| Name | Summary |
|---|---|
| Controlled experiment | Randomize the input to be acted on, then analyze again. Randomization breaks confounding, so any change in the target is attributable to the input. |
| Natural experiment | Find a setting where the input varied for reasons unrelated to the target, an instrumental variable. Variation of that kind supports causal inference. |
| Causal graph | Write out what plausibly causes what, then check whether the model's terms are consistent with the conditional independences the graph implies. |
| Control for confounders | If a third variable plausibly moves both the input and the target, stratify by it or condition on it before training again. Gaps in the Quality by Input Range chapter are a clue. |
Evidence
Every chapter of the analysis in full: the figures the story rests on, and everything else the analysis measured.
1 Headline
Diafunc modeled Target from 4 columns, of which the model uses 3. The model explains 85% of Target and relies most on In4, with which the model's output rises.
- One target was modeled; 98 rows were held out for checking.
- A Poisson regression on the numeric inputs had a mean Poisson deviance of 0.0576; the model added -16 points of Poisson deviance removed, so did not beat the baseline.
- The checks found a Critical concern: data has 65% exact duplicate rows, so the held-out split is likely contaminated (98 rows checked).
- Best R²
- 0.8483
- Best quality
- 83.92%
- Best target
- target
- Targets
- 1
2 Concerns
| Severity | Section | Concern | Detail |
|---|---|---|---|
| Critical | Data | Data has 65% exact duplicate rows: the held-out split is likely contaminated | 317 of 488 rows (65%) are exact duplicates of at least one other row (largest duplicate group: 11 identical rows). At this density some duplicate pairs very likely straddle the held-out split, in which case the held-out R² or accuracy overstates how the model does on rows it has not seen. Deduplicate before deploying: drop the duplicates outright, or, where they encode intentional row weights, aggregate them into one weighted row. An analysis of the deduplicated data reads the headline metric without the inflation. |
| Warning | Model | The model does not beat a Poisson regression on the numeric inputs on Target | On the held-out rows the model leaves 16% more Poisson deviance than a Poisson regression on the numeric inputs. The interval of the margin includes zero: on this many held-out rows the two are not told apart. The formula adds little over the reference predictor; whether it is worth its complexity depends on the use. A plain regression on the same inputs does as well or better, so the inputs do carry signal; the search did not find a better formula in the time it had. Row by row the two are about level. |
How to read this
Data-quality, model-quality, and deployment-risk issues detected during the analysis appear below, ordered by severity. Critical concerns are listed first; Warning and Info entries follow.
3 Discovered Relations
| Target | Formula | Quality | Complexity | Error |
|---|---|---|---|---|
| Target | Target = -1.6 + 0.37 × In2 + 0.44 × In3 + 0.45 × In4 | 83.92% | 12.2% | 15.1% |
How to read this
One model per target, each a formula with its quality, complexity and error. Higher quality at lower complexity describes the data most clearly. The findings below read the models together; the per-target chapters in the Model section carry each formula, its predictions and its diagnostics.
4 Model
The model's output relies most on in4, accounting for 71% of its reliance on inputs. Its quality on the held-out rows is measured by a share of variation explained value of 0.848. A warning is flagged: the Concerns chapter describes that the model does not surpass a Poisson regression baseline.
Most notable in the Model section:
In4carries most of the model (71% of its reliance on inputs)
How to read this
The Model section presents one model per target: its formula, its predictions on held-out rows, the inputs it relies on, the baselines it is measured against, and diagnostics of how it behaves. With more than one target, the Composite Model, Multi-Output and Leaderboard chapters read the models as a set.
4.1 Target
The Model section builds one model for Target: a formula that expresses it through the other columns. Diafunc searches the space of mathematical expressions for the formula that balances accuracy against simplicity, so the result is not a black box but an expression that can be read, evaluated and questioned. The chapters that follow show the formula, score it on rows it never saw during training, and take it apart term by term.
4.1.1 Target Model
The model's output rises with In4, the strongest input, and also rises with In3 and In2. The model relies on In4 for 71% of the explained variation, with an overall training quality of 83%.
Target = -1.6 + 0.37 × In2 + 0.44 × In3 + 0.45 × In4| Value | Role | Input | Unit |
|---|---|---|---|
| -1.6 | Intercept | Unit of the model's output | |
| 0.37 | Slope | In2 | Output units per unit of In2 |
| 0.44 | Slope | In3 | Output units per unit of In3 |
| 0.45 | Slope | In4 | Output units per unit of In4 |
- Share of the output's variation they carry
- 100%
- Rows the search trained on, measured
- 195
- Rows the search trained on
- 195
- Held out R²
- 0.8483 (95% CI 0.79 – 0.8887)
- Constant base
- -1.6
- Constants in the formula
- 4
- Terms that reach 95% of the output's variation
- 3
- Terms that read an input, constants multiplied in
- 3
- Training rows
- 390
- Held-out rows
- 98
- Inputs the formula reads
- 3
- Terms in the formula
- 4
- Inputs the formula does not read
- In1
How to read this
The three gauges score the model on the held-out rows: rows that were set aside before the search and that the model never saw. They are the figures to quote for how the model does on new data.
| Metric | Meaning |
|---|---|
| Quality | How well the model predicts on the held-out rows, marked down for the size of its formula: 100% is an exact prediction, 0% is no better than always predicting the average. It is not R², and sits at or just below it |
| Error | How far the predictions lie from the held-out values, as a share of the miss of always predicting the average: 0% is an exact prediction, 100% is no better than the average |
| Complexity | The size of the formula as a share of the complexity the search was allowed: a low figure is a short formula |
The figures under During the search are the same measures on the training rows. They are what the search ranked its candidates by, so they describe the search and tend to flatter the model it chose. The Generalization chapter sets the training rows beside the held-out rows, and the Baselines chapter sets the model beside its baselines. The time and credit gauges account for the computation that produced the model, not for the model itself.
4.1.2 Prediction
The model predicts Target for any row that supplies the other columns. This chapter applies it to the held-out rows, the rows set aside before training, so that the predictions can be compared with values the model never saw. The rows themselves and the function that predicts them are kept as reference under Foundation, with the processed data.
4.1.2.1 Prediction Chart
On the held-out rows the model explains 85% of the variation in the counts (R² 0.848); its mean absolute error is 0.428. Mean Poisson deviance 0.0666, with under-dispersion (Pearson ratio 0.0665 < 1). The counts vary less than Poisson expects, often a bounded or capped process. RMSE (root mean squared error) on counts is 0.582.
- Dispersion
- 0.06647 (95% CI 0.04709 – 0.08675)
- Error
- 15.1% (95% CI 10.91% – 20.66%)
- MAE
- 0.4284 (95% CI 0.355 – 0.5035)
- N
- 98
- Training rows
- 390
- Poisson deviance
- 0.06662 (95% CI 0.0466 – 0.08695)
- Quality
- 83.92% (95% CI 78% – 88.38%)
- R²
- 0.8483 (95% CI 0.79 – 0.8887)
- RMSE
- 0.5824 (95% CI 0.4774 – 0.6739)
- Target
- Target
How to read this
Actual Target on the horizontal axis against the model's predicted value on the vertical axis, for the held-out rows the model never saw during training. Points on the diagonal are perfect predictions; systematic vertical offsets reveal bias, vertical spread reveals noise, and points far from the diagonal are rows the model handles poorly. With 10,000 held-out rows or more, the rows are counted instead of drawn one point each: each cell is a range of the actual and a range of the predicted value, its color the number of rows that fall in both, and the cells whose two ranges match hold the rows the model predicts well.
4.1.2.2 Held-out Dispersion
- Dispersion
- 0.06647
- N
- 98
- Observed mean
- 5.224
- Observed variance
- 2.235
- Predicted mean
- 5.181
- Verdict
- under-dispersed
How to read this
Whether the model reproduces the amount of variation the held-out counts of Target actually show. The observed and predicted means say whether the overall level is right; the dispersion ratio compares the observed variance against the variance a Poisson process at that level would produce. A ratio near one means the counts vary as much as such a process, a ratio well above one means they vary more (overdispersion, so any prediction interval is too narrow), and a ratio well below one means they vary less.
4.1.3 Which inputs matter most
The model's output rises with In4 on every one of the 1,170 steps between observed values. The model's output rises with In3 on every one of the 1,170 steps between observed values. The model's output rises with In2 on every one of the 1,560 steps between observed values. Moving In4 from 5.00 to 6.00, its first and third quartile, with every other input as observed raises the model's output by 0.45 on average. These are measurements on the rows the search trained on, inside the observed ranges; outside them nothing is claimed.
| Input | Share of importance | Permutation importance | Sensitivity | Residual correlation | Occurrences | Model's output as the input rises | Change, first to third quartile |
|---|---|---|---|---|---|---|---|
| In4 | 70.85% | 0.7953 | 0.45 | -0.2965 | 1 | Rises (1170 of 1170 steps) | +0.45 (5 to 6) |
| In3 | 21.22% | 0.2382 | 0.44 | 0.06217 | 1 | Rises (1170 of 1170 steps) | +0.44 (5 to 6) |
| In2 | 7.93% | 0.0890 | 0.37 | 0.3862 | 1 | Rises (1560 of 1560 steps) | +0.74 (4 to 6) |
- Rows the search trained on, measured
- 195
- Inputs the formula does not read
- In1
How to read this
The chart ranks the inputs of the Target model by permutation importance: how much the model's error rises when that input's values are shuffled across rows so that it carries no information. A longer bar means the model relies on the input more. The order says which columns the model reads most, and those are the columns where measurement quality and data collection matter most. A short bar marks an input the model barely uses, which is a fact about this model and not a verdict on the column. Importance describes the model's reliance and nothing more: an input can rank high because it stands in for something the data does not contain. Where the search measured it, an input also carries the direction in which the model's output moves as the input rises, counted on the rows the search trained on, and the change of the output when the input alone goes from its first to its third quartile; both describe the formula inside the observed range and say nothing about causes.
4.1.4 Baselines
The model does not beat the simple yardsticks: a Poisson regression on the numeric inputs is ahead by 16% of its Poisson deviance on the held-out rows. The model's complexity is not buying anything yet. A plain regression on the same inputs does as well or better, so the inputs do carry signal; the search did not find a better formula in the time it had. Of the 98 held-out rows where the two differ, the model is closer on 49% (95% interval 39% to 59%). The interval of the margin includes zero: on this many held-out rows the two are not told apart.
- Best yardstick's score
- 0.05761 (95% CI 0.03995 – 0.07746)
- Effect size d
- -0.0889 (95% CI -0.2698 – 0.1074)
- Effect size d magnitude
- negligible
- Margin over the best yardstick
- -0.1564 (95% CI -0.4778 – 0.0851)
- Margin's interval stays above zero
- No
- Model
- 0.06662 (95% CI 0.0466 – 0.08695)
- Held-out rows
- 98
- Held-out rows without an answer
- 0
- Target
- Target
- Rows where the model is closer
- 48.98% (95% CI 39.31% – 58.73%)
- Rows compared
- 98
- Rows level
- 0
- Regression yardstick trained with gaps filled (too few complete rows)
- No
- Training rows of the regression yardstick
- 390
How to read this
The table scores the model for Target against simple yardsticks on the same held-out rows. A yardstick is a predictor simple enough to need no explanation: a constant (the average, the middle value, the most common class or level, the average count), a plain regression on the numeric inputs (a straight line, a logistic regression, a Poisson regression), and for a series over time a forecast from its own past (the last value, the value one season back, the recent trend extended). Which of them appear depends on what kind of column Target is. They are yardsticks only: the analysis has one model, and it is the last row. A regression yardstick answers every row the model answers: where one of its numeric inputs is empty, it reads the average of that input over the training rows. Nothing is filled in for the model. A yardstick whose row is empty could not be scored, and the summary then says what the margin is measured against.
Read the columns from top to bottom. Each kind of target is scored by the measure it is judged on: R² and MAE for a continuous value, with MASE added for a series (below 1 means a typical miss is smaller than the series' usual step from one row to the next); accuracy and balanced accuracy for a yes/no value, with the Brier score where a probability is predicted; accuracy and macro F1 for a class; weighted kappa and the distance in levels for an ordered level; Poisson deviance and RMSE for a count. When the model clears every yardstick by a wide margin, its structure earns its place. When it is level with the regression yardstick, a plain regression describes the data as well. When it is level with the constant, the inputs carry no usable signal for Target.
The margin over the best yardstick comes with an interval. When that interval includes zero, the lead is not settled on this many held-out rows. The share of rows on which the model is closer than the best yardstick is counted only on the rows where the two differ; for a yes/no value or a class these are the rows where the two disagree.
4.1.5 Generalization
The model explains held-out rows about as well as training rows (held-out R² 0.848, gap between training and held-out R² -0.0152): it generalizes. The structure the formula encodes carries from the training rows to the held-out rows.
- Generalization gap
- -0.01517
- N test
- 98
- Target
- Target
- Test error
- 0.4284 (95% CI 0.355 – 0.5035)
- Test R²
- 0.8483 (95% CI 0.79 – 0.8887)
- Train error
- 0.1493
- Train R²
- 0.8331
- Verdict
- Generalizing
How to read this
The chart sets the score of the Target model on the rows it trained on beside its score on the held-out rows it never saw, and the gap between the two bars is the finding. The score is the one that suits the target: R² for a quantity, balanced accuracy for a yes/no target, weighted kappa for ordered levels. The held-out figure carries a bootstrap 95% confidence interval, which says how far it could move with a different held-out draw; the training bar has none, because it is measured on the rows that shaped the model.
The verdict reads from the two figures. Generalizing: the held-out score is strong, or training and held-out agree closely; the structure the formula encodes carries to new rows. Overfit: training is far ahead of held-out and the held-out score is not strong on its own; the formula has absorbed noise from the training rows, and the training figure overstates what to expect on new data. Underfit or noisy: held-out is far ahead of training; this is not strength, it usually means an outlier or a flat region in the training rows that the held-out rows happen to avoid. Insufficient: fewer than 30 held-out rows and no strong held-out score, too little evidence for a negative verdict. A model that predicts the same value for every held-out row gets no verdict: there is nothing that could carry over. In every case the width of the held-out interval says how much weight the gap can bear.
4.1.6 How each term responds to its inputs
Read as a sum, the formula has 4 additive terms: 3 that read an input and a constant base. All that read an input are needed to reach 100% of the variation of the model's output on the 195 rows the search trained on: the largest are 0.37 × In2 (39%), 0.44 × In3 (36%) and 0.45 × In4 (25%). A share describes the model's output, not the target. Each curve below shows one term's response to its free inputs, with the other inputs held at the training median.
- Additive
- Yes
- Rows the search trained on, measured
- 195
- Terms in the formula
- 4
- Notation
- -1.6 + 0.37 × In2 + 0.44 × In3 + 0.45 × In4
- Target
- Target
How to read this
The formula for Target is split into its additive terms, and each term is traced as one of its inputs moves across the training range while the other inputs stay at their medians. Each curve is therefore a profile of the term at the typical row, not an average over the data: it shows what the term contributes at every value of that input when everything else is ordinary.
A term that depends on several inputs gets one curve per input; a constant term gets none. Terms the formula subtracts are drawn with their sign, so a curve below zero pulls the prediction down. At any fixed row, the terms add up to the model's prediction. Where the search measured it, a term also carries its share of the variation of the model's output on the rows the search trained on; that is a share of what the formula returns, not of Target itself. The notebook lets the reader trim each range to the central 95% of the input. When the formula has no additive structure, the chapter traces the whole formula as a single term along each of its inputs, and the Sensitivity chapter is the better read for how the inputs combine.
4.1.7 How the formula behaves outside the data
The model diverges beyond the observed range of in2, in3, in4: predictions there are not supported by the formula's shape. Extrapolate with caution.
- Bounded at infinity
- No
- Rows the search trained on, measured
- 195
- Inputs the formula reads
- 3
- N singularities
- 0
- Risky inputs
- in2, in3, in4
- Target
- Target
| Input | Kind |
|---|---|
| In2 | Negative infinity |
| In3 | Negative infinity |
| In4 | Negative infinity |
| Input | Kind |
|---|---|
| In2 | Positive infinity |
| In3 | Positive infinity |
| In4 | Positive infinity |
How to read this
The formula for Target is read as a mathematical object, without data, to state how it behaves beyond the range it was trained on. A black box cannot answer that question; a formula can. For each input the chapter reports the limit as the input grows without bound in either direction (whether the formula diverges, settles to a finite value or oscillates, with the sign tracked exactly, so that a term like minus two times the cube of the input is reported as diverging downward on the right and upward on the left), whether the formula stays finite in every direction, and any singularity with its solved location: a division whose denominator reaches zero, a logarithm or square root whose argument leaves its domain, or a periodic pole.
The inputs along which the formula diverges or meets a singularity are listed as the risky ones; along them the formula's shape gives no support for extrapolation. A formula that stays finite everywhere with no singularity carries no structural hazard when extrapolated; the data still ends where it ends, and a prediction outside the training range remains unsupported by evidence. The plain-language summary at the end states the same facts in one paragraph.
4.1.8 Contributions per term
- Additive
- Yes
- N rows
- 5
The notebook draws this chapter's figure from each held-out row; the row-by-row detail is there.
How to read this
For up to five representative held-out rows, the chapter draws one bar chart showing how each additive term of the formula builds that row's prediction of Target. The five rows are the worst prediction (largest absolute residual), the best (smallest), the median miss, the row with the smallest prediction and the row with the largest. When two roles fall on the same row, which is common with few held-out rows, the earlier role keeps it and fewer than five charts appear.
In each chart the bars stack the terms' contributions at the row's input values, terms the formula subtracts sit below zero, and the bars sum to the prediction. The dashed line marks the actual value; the distance between it and the top of the stack is the residual. The bars are evaluated at the same precision as the predictions, so they agree with the Prediction Chart. When the formula has no additive structure, each chart holds one bar for the whole formula, and the Sensitivity chapter is the better read for how individual inputs move it.
4.1.9 What each term is worth
Full-formula R² is 0.848. Each bar below shows how much R² drops when that term is replaced by its mean contribution. Large drop: load-bearing term; near zero: decoration the formula could lose; negative: overfit signal (ablation improves R²).
- Additive
- Yes
- Baseline R²
- 0.8483
- Held-out rows
- 98
The notebook draws this chapter's figure from each held-out row; the row-by-row detail is there.
How to read this
For each additive term of the formula for Target, the chart shows how much the held-out R² drops when that term is replaced by its mean contribution, so that it no longer varies from row to row. The bars are sorted from the largest drop down, and the dashed line marks the R² of the full model.
A large drop means the term carries real predictive work; the complexity it adds is paid for. A small drop means the term adds little beyond its average value: it may refine a stronger term, or its curve in the chapter "How each term responds to its inputs" may be nearly flat. A negative drop means the held-out R² rises without the term, which is a sign that the term tracks noise in the training rows. Each figure describes what a term contributes inside the model as discovered; the chart diagnoses the model, it is not a guide to removing terms. When the formula has no additive structure, the chart shows one bar for the whole formula.
4.1.10 What-If
| Item | Value |
|---|---|
| In2 | 5 |
| In3 | 5 |
| In4 | 6 |
| Term 1 adds | -1.6 |
| Term 2 adds | 1.85 |
| Term 3 adds | 2.2 |
| Term 4 adds | 2.7 |
| Prediction | 5.15 |
- Additive
- Yes
- Held-out rows for the interval
- 98
- Terms in the formula
- 4
- Residual P5
- -0.99
- Residual P95
- 0.926
- Residual std
- 0.5837
How to read this
The chapter shows the model's prediction for Target at one input combination, with the contribution of each term stacked beneath it. The inputs start at their training medians, so the first view is the prediction for a typical row, and each input can move within the range the training rows cover; the notebook lets the reader change the inputs and watch the prediction and the stack update.
In the stack, a long bar marks the term that dominates at this input combination, often the one whose removal costs most in the chapter "What each term is worth". A short bar marks a term that makes a small adjustment here; the chapter "How each term responds to its inputs" shows whether it stays small everywhere or grows toward the edges of the range. A negative bar marks a term that pulls the prediction down at this point, either because the formula subtracts it or because it evaluates below zero.
The inputs stay inside the training range for a reason: the formula computes a value anywhere, but outside that range no training row supports it. When the formula has no additive structure, the stack holds a single bar for the whole formula and the inputs move the prediction directly.
4.1.11 Sensitivity
The prediction responds to 3 inputs. The formula is additive: each curve below shows the change in the prediction per unit of one input (∂ŷ/∂xᵢ) independently. Flat curves mean a constant linear response, bent curves a non-linear response that depends on the position in the range, and zero crossings a response that reverses sign between regions.
- Additive
- Yes
- Terms in the formula
- 4
How to read this
For each input, the chart traces the rate at which the predicted Target changes as that input moves across its training range, with the other inputs held at their medians. The rate is the local slope of the prediction with respect to the input, measured by a small finite difference at each point. A table summarizes each input with the mean, minimum and maximum of the absolute slope, which ranks the inputs by how strongly the prediction responds to them.
A flat curve means the model is close to linear in that input: the prediction responds at the same rate everywhere in the observed range, which is easy to state and says nothing about values beyond that range. A curve that grows or shrinks marks diminishing or accelerating returns, and whether that is plausible is a question for the domain. A curve that crosses zero means the prediction rises with the input in one part of the range and falls in another, which is either a real optimum or a sign that the formula bends around a few rows. A curve with a large magnitude marks an input to which the prediction is highly responsive; the chapter "What each term is worth" says whether the term behind it also carries predictive work.
In the table, the mean absolute slope ranks the inputs by typical responsiveness. A minimum close to the maximum means the responsiveness is nearly constant; a wide gap means it varies strongly across the range. The chapter reads the whole formula, so it applies whether or not the formula splits into terms.
4.1.12 How two inputs act together
In2, in3 and in4 enter linearly. No term reads two inputs, so the grid below renders empty; each input's contribution can be read independently in the chapter "How each term responds to its inputs".
- Additive
- Yes
- Inputs in different terms act separately
- Yes
- Rows the search trained on, measured
- 195
- Inputs the formula is linear in
- In2, In3, In4
- Terms in the formula
- 4
- Input pairs not tested
- 0
How to read this
When a single term of the formula combines two or more inputs, the prediction's response to one of them depends on the value of the other. This chapter draws one surface for each such pair: the predicted Target across the plane of the two inputs, with every other input held at its training median. Pairs that never share a term have no interaction by construction, and the curves in the chapter "How each term responds to its inputs" already say everything about them. Where the search tested a pair, the summary says on how many pairs of rows the two inputs acted together in the model's output; that is a statement about the formula, not a finding that the data hold no other interaction.
Parallel stripes mean the two inputs act almost independently, so the surface is close to a sum of two one-input curves. Curved contours mean a real interaction, where the slope along one input changes with the other; the location of the peak or trough matters. A saddle, rising along one axis and falling along the other, marks a trade-off between the two. A sharp edge or cliff means the formula encodes a change of regime, and the curve for the same input in the chapter "How each term responds to its inputs" shows whether the bend appears there too.
The grid is deliberately coarse, and the color scale runs from low to high predictions. When the formula has no additive structure, the single whole-formula term holds every input, so every pair appears; the number of surfaces is capped for models with many inputs.
4.1.13 Worst Predictions
The worst held-out row was missed by 1.78 (target's scale). The 10 rows with the largest absolute residuals out of 98 held-out rows are shown. Each card breaks the failure down into per-term contributions and input values, so the failure mode is inspectable. A large residual with one concentrated term contribution means one term got this row wrong; a large residual with balanced contributions means a structural mismatch.
| Actual | Predicted | Residual | In2 | In3 | In4 |
|---|---|---|---|---|---|
| 9.000 | 7.220 | +1.780 | 7 | 7 | 7 |
| 5.000 | 6.770 | -1.770 | 7 | 7 | 6 |
| 9.000 | 7.300 | +1.700 | 6 | 7 | 8 |
| 7.000 | 5.600 | +1.400 | 5 | 5 | 7 |
| 3.000 | 4.260 | -1.260 | 5 | 4 | 5 |
| 4.000 | 5.220 | -1.220 | 4 | 6 | 6 |
| 5.000 | 3.890 | +1.110 | 4 | 4 | 5 |
| 4.000 | 5.060 | -1.060 | 6 | 6 | 4 |
| 4.000 | 4.990 | -0.990 | 7 | 5 | 4 |
| 2.000 | 2.990 | -0.990 | 4 | 4 | 3 |
- Additive
- Yes
- N extrapolated
- 0
- Held-out rows
- 98
- Held-out rows for the interval
- 98
- Residual P5
- -0.99
- Residual P95
- 0.926
- Residual std
- 0.5837
How to read this
The held-out rows on which the model misses Target by the most, sorted by absolute residual with the largest first and capped at ten. Each card shows the row's actual value, the prediction, the residual, the contribution of every additive term at that row and the row's input values. R² and MAE say how good the model is on average; these cards say where it fails and let each failure be read input by input and term by term.
A large residual with one dominant term means that term produced the wrong value at this row; the row's value for that input, read against its curve in the chapter "How each term responds to its inputs", shows whether it fell in a region the formula models poorly. A large residual with balanced contributions means every term was a little off and the sum still missed, which points to a shape the formula cannot represent in that region. Residuals that change sign from card to card mean the misses are not a consistent bias but scattered errors on particular rows. When the formula has no additive structure, each card shows a single whole-formula contribution.
4.1.14 Count Distribution
Predicted against observed counts across 98 held-out rows: under-dispersed (Pearson ratio 0.0665). The observed counts cluster more tightly than Poisson expects, often a bounded or capped count process.
- Bin width
- 1
- Binning
- integer
- Dispersion
- 0.06647
- N
- 98
- Observed mean
- 5.224
- Predicted mean
- 5.181
- Target
- Target
- Verdict
- under-dispersed
- Zero observed
- 0
- Zero predicted
- 0
| Count | Lower | Observed | Predicted | Upper |
|---|---|---|---|---|
| 0 | 0 | 0 | 0 | 0 |
| 1 | 1 | 0 | 0 | 1 |
| 2 | 2 | 2 | 1 | 2 |
| 3 | 3 | 13 | 13 | 3 |
| 4 | 4 | 16 | 16 | 4 |
| 5 | 5 | 21 | 20 | 5 |
| 6 | 6 | 29 | 34 | 6 |
| 7 | 7 | 13 | 12 | 7 |
| 8 | 8 | 1 | 2 | 8 |
| 9 | 9 | 3 | 0 | 9 |
How to read this
The chart sets the held-out counts of Target beside the model's predictions rounded to the nearest non-negative integer, one pair of bars per count value, with the last bin collecting the high counts. Where the two distributions diverge, the model misstates how the counts arise; this is the check that tells a count target apart from a continuous one.
A long tail of observed counts the predicted bars do not reach, or a tall observed spike at zero, means the counts vary more than the model reproduces: overdispersion, or an excess of zeros. The dispersion ratio, also reported with the held-out distribution figures in the Prediction chapter, compares the observed variance with the variance a Poisson process at the same mean would show. A ratio near one means the counts vary as much as such a process, well above one means they vary more, and well below one means they vary less.
4.1.15 Search Progress
The search ran for 44 generations, ending at quality 0.825 (gain +0.179 from generation 2). The curve shows a substantial climb. The last improvement came after 100% of the search time: the search was still improving when it stopped, so a longer time limit is likely to find a better model.
- First generation
- 2
- First quality
- 64.66%
- Last generation
- 45
- Last improvement ms
- 2,529,039
- Last quality
- 82.51%
- Max quality
- 82.51%
- N points
- 44
- N total generations
- 44
- Search time ms
- 2,529,039
- Stop reason
- Out of budget
How to read this
The line traces the best quality the search for the Target model had reached at each generation, from the first generation to the last. Quality is the technology's combined score on the training rows, the figure the Model Metrics chapter defines. A curve that flattens well before the last generation means the search converged: the budget was more than enough, and the model is the one the search settled on. A curve still rising at the last generation means the budget ended before the search settled, and the model is the best the search reached within it. The chart describes the search, not the data: it says whether the search had time to settle, not whether more rows would help.
4.1.16 Model Constraints
Across 98 held-out predictions, the formula is mathematically well-behaved: every output is finite, stays inside the training-target ±2σ band, and no prediction runs away beyond the bounded-prediction limit. It is safe to consume downstream at face value.
| Name | Status | Detail |
|---|---|---|
| Output coherent | Pass | no non-finite predictions across 98 rows |
| Within domain | Pass | no rows outside the training target's ±2σ range across 98 rows |
| Bounded predictions | Pass | no runaway predictions beyond ±10σ across 98 rows |
- Held-out rows
- 98
- N non finite
- 0
- N outside domain
- 0
- N runaway
- 0
- Verdict
- Clean
How to read this
Sanity checks on the Target formula, run against the held-out rows. Three checks: every prediction is a finite number (output coherent), predictions stay within the range of the training target widened by two standard deviations on each side (within domain), and no single prediction runs away more than ten standard deviations from the training mean (bounded predictions).
These are pass or fail guards, not quality measures: a formula can clear all three and still predict poorly. A failure is the stronger signal. It means the formula misbehaves numerically somewhere in the data it was shown, and any downstream use of it should wait until that is understood.
4.1.17 Modeling
Measured during the search, on the rows the model was trained on
- MAE of a straight line through the inputs (same rows)
- 0.3871
- RMSE of a straight line through the inputs (same rows)
- 0.4988
- Training MAE (rows the search trained on)
- 0.4379
- MAE of always predicting the mean (same rows)
- 1.111
- RMSE of always predicting the mean (same rows)
- 1.413
- Training R² (rows the search trained on)
- 0.8471
- Training RMSE (rows the search trained on)
- 0.5526
- Rows the search trained on, measured
- 195
- Error
- 14.93%
- Inputs the straight line used
- 4
- N train
- 390
- Quality
- 82.53%
How to read this
The function below trains the model for Target on the processed data. Its output is the model that every other chapter of this section scores and dissects.
4.2 Findings
1 finding from the model diagnostics (1 Warning) is listed in the Concerns chapter at the top of the analysis.
5 Data
Multiple variables track together in this data, with in2 having 10 levels measured in 488 rows and no variables missing. Substantial exact duplicates are present at 65%, which is flagged as critical in the Concerns chapter.
Most notable in the Data section:
In3andTargetmove together (correlation 0.823)In1andIn2move together (correlation 0.763)In4andTargetmove together (correlation 0.751)
How to read this
The Data section describes the table as supplied, before any modeling: the Column Profiles overview of every column and its role, one chapter per column with its distribution and correlations, a closer look at columns with notable characteristics, Duplicate Rows, Leakage Suspicion, a map for paired coordinate columns, and the General chapter with descriptive statistics, correlations, missing values, outlier ranges, clusters and segments across all columns.
5.1 Column Profiles
Profiled 5 columns across 488 rows (5 numeric, 0 non-numeric): every column is at least 75% populated. Distributional statistics on numeric columns can be read at face value.
| Name | Role | Type | N | Missing | Mean | Std | Min | P25 | Median | P75 | Max | Skewness | CV |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Target | Target | integer | 488 | 0 | 5.230 | 1.422 | 1 | 4 | 5 | 6 | 9 | -0.1002 | 0.2719 |
| In1 | Input | integer | 488 | 0 | 5.506 | 1.526 | 0 | 4.750 | 6 | 6 | 9 | -0.1544 | 0.2771 |
| In2 | Input | integer | 488 | 0 | 5.084 | 1.551 | 0 | 4 | 5 | 6 | 9 | -0.1904 | 0.3051 |
| In3 | Input | integer | 488 | 0 | 5.344 | 1.136 | 2 | 5 | 5 | 6 | 8 | -0.4598 | 0.2125 |
| In4 | Input | integer | 488 | 0 | 5.611 | 0.937 | 2 | 5 | 6 | 6 | 8 | -0.3835 | 0.1671 |
- N columns
- 5
- N high missing
- 0
- N rows
- 488
- N strongly skewed
- 0
- Table branch
- main
- Table ID
- 5ca0afe1-6393-477c-bddc-5c09ece16485
How to read this
One row per column of the analyzed table: its role in the analysis (the target being predicted, an input it is predicted from, or the segment column), its type, value count and missing count, and for numeric columns the mean, standard deviation, minimum, quartiles, maximum and interquartile range. A column with 25% or more missing values is flagged; its mean and quartiles describe only the rows that have a value.
Two further figures describe the shape of a numeric column. The skewness says whether the values lean to one side: close to zero the column is symmetric, from 0.5 it is moderately skewed and from 1 strongly skewed, with the longer tail on the high side when the figure is positive and on the low side when it is negative. In a strongly skewed column the mean is pulled toward the tail, and the median is the better guide to a typical value. The coefficient of variation is the standard deviation divided by the mean: it compares how widely columns spread regardless of their units, and is given only for a column whose mean is positive.
These are the columns as supplied, before any modeling: the reference against which every later chapter reads.
5.2 In1
In1: 0 to 9, median 6, skewed to low values, 10 distinct values; moves with In2 most (r 0.763).
How to read this
A column chapter profiles one column: its distribution in figures and charts, and how it relates to the other columns. It opens with one line on what the column holds. The Description chapter carries its figures, with every other column summarized for each of its values where it has few, and its association with every other column where it is numeric. The Visualization chapter carries its charts. A table with a date column adds an Over Time and a Time Series chapter; a target with classes adds a Class Balance chapter.
5.2.1 Description
| Statistic | Value |
|---|---|
| Type | integer |
| Values present | 488 |
| Distinct values | 10 |
| Continuous | yes |
| Decimal places | 0 |
| Minimum | 0 |
| Maximum | 9 |
| Mean | 5.506 |
| Median | 6 |
| Mode | 6 |
The first 10 rows of 17. The other 7 are in the notebook.
How to read this
The figures below summarize In1 in numbers: where its values center (mean, median, mode), how far they spread (range, variance, standard deviation, interquartile range) and how they are distributed across percentiles. They are the reference for every chart in the Visualization chapter and for the correlation measures further down.
The figures for In1 are grouped: the column itself first, then where its values center, how far they spread, the shape of their distribution and its percentiles. Each statistic is explained in the Description chapter under General in the Data section.
5.2.1.1 By Value
In1 accounts for 57% of the spread in Target: 9 averages 4.60 above 2 (95% interval 3.58 to 5.63). An association, not a cause.
| Target | Value | Rows | Average or share | 95% interval |
|---|---|---|---|---|
| Target | 9 | 7 | 7.143 | 6.154 to 8.132 |
| Target | 8 | 48 | 6.958 | 6.647 to 7.269 |
| Target | 7 | 58 | 6.276 | 6.012 to 6.54 |
| Target | 6 | 140 | 5.65 | 5.491 to 5.809 |
| Target | 5 | 113 | 4.938 | 4.77 to 5.106 |
| Target | 4 | 81 | 3.975 | 3.79 to 4.16 |
| Target | 3 | 26 | 3.615 | 3.271 to 3.96 |
| Target | 2 | 13 | 2.538 | 2.069 to 3.008 |
| In1 | Rows | In2 | In4 | In3 | Target |
|---|---|---|---|---|---|
| 0 | 1 | 5 | 2 | 4 | |
| 1 | 1 | 2 | 5 | 3 | 2 |
| 2 | 13 | 1.923 | 4.462 | 3.231 | 2.538 |
| 3 | 26 | 3.269 | 4.923 | 4.269 | 3.615 |
| 4 | 81 | 3.765 | 5.16 | 4.506 | 3.975 |
| 5 | 113 | 4.761 | 5.478 | 5.177 | 4.938 |
| 6 | 140 | 5.621 | 5.8 | 5.679 | 5.65 |
| 7 | 58 | 6.224 | 6.017 | 6.052 | 6.276 |
| 8 | 48 | 6.833 | 6.229 | 6.458 | 6.958 |
| 9 | 7 | 7.143 | 6.429 | 6.286 | 7.143 |
How to read this
The rows are grouped by the value of In1, and every other column is summarized within each group with the same descriptive statistics. A statistic that shifts from group to group marks a column that moves with In1; one that stays level marks a column that does not. Groups with few rows produce unstable figures, so the row count of each group is the first thing to read. Where the groups differ in a target by more than chance would give, the chapter states by how much; it states that the two go together, not that one brings the other about.
5.2.1.2 Correlation
| Column | Pair of | Association | Pearson's r | 95% CI (Pearson) | Spearman's ρ | Kendall's τ | q-value | Rows |
|---|---|---|---|---|---|---|---|---|
| In2 | two numbers | 0.7628 | 0.7628 | 0.723 to 0.7976 | 0.7492 | 0.6452 | < 0.001 | 488 |
| Target | two numbers | 0.7531 | 0.7499 | 0.7083 to 0.7864 | 0.7531 | 0.655 | < 0.001 | 488 |
| In3 | two numbers | 0.6493 | 0.6493 | 0.5948 to 0.6978 | 0.6343 | 0.5388 | < 0.001 | 488 |
| In4 | two numbers | 0.4438 | 0.4423 | 0.368 to 0.511 | 0.4438 | 0.3743 | < 0.001 | 488 |
How to read this
Each row pairs In1 with one other column, numeric or categorical, and reports how strongly the two go together on an association from 0 to 1, and whether an association this strong could be chance. For another numeric column the row adds the direction: Pearson's coefficient measures a straight-line relationship, and Spearman's rho and Kendall's tau measure a monotonic relationship on ranks, which makes them resistant to outliers and lets them see a curved but consistently rising or falling pattern. For a categorical column the association is the correlation ratio η, which has no direction. A strong association is not an explanation: two columns can go together when a third one moves both.
The table lists every other numeric or categorical column against In1, with the pair's kind and association, three correlation coefficients and a confidence interval where the other column is numeric, the q-value and the rows the pair was measured on.
| Measure | Meaning |
|---|---|
| Kind | What the other column holds: a number (numeric) or categories (numeric_categorical). A categorical column is one that is not numeric and holds between 2 and 50 distinct values |
| Association | The strength of the pair from 0 to 1. With a numeric column: the stronger of Pearson's coefficient and Spearman's ρ, without its sign. With a categorical column: the correlation ratio η, the square root of the share of In1's spread that lies between the categories |
| Covariance | Numeric partner only. The direction of the relationship (positive or negative) in the units of the two columns. It is not standardized, so its size says little about strength without the columns' own spread |
| Pearson's Correlation Coefficient | Numeric partner only. The strength and direction of a linear relationship, from -1 (a perfect falling line) through 0 (no linear relationship) to 1 (a perfect rising line) |
| Spearman's ρ | Numeric partner only. The strength and direction of a monotonic relationship, computed on ranks rather than values, from -1 to 1. It follows a curved pattern as long as it rises or falls consistently, and outliers move it little |
| Kendall's τ | Numeric partner only. The strength and direction of association from the agreement and disagreement of rank order across pairs of rows, from -1 to 1. It is the steadier of the two rank measures when the rows are few |
| q-value | How likely a pair this strong is among unrelated columns, adjusted for the number of pairs involving In1 (Benjamini-Hochberg). Among all pairs reported with q below 0.05, about 5% or fewer are expected to be false discoveries |
| 95% CI (Pearson) | Numeric partner only. The lower and upper bounds of the 95% confidence interval on the Pearson coefficient, from the Fisher z-transform. An interval that contains 0 leaves the direction of the relationship open |
| Rows | The rows in which both columns hold a value. The q-value and the interval of a pair rest on this count |
5.2.2 Visualization
The rows of In1 are not charted in their order: consecutive rows are no more alike than rows picked at random (a correlation of -0.02 between each value and the next), so the order carries no pattern to show.
How to read this
The charts in this chapter were chosen from the type of In1 and of the columns it is compared with: a numeric column gets a histogram, a box plot and, when its row order carries a pattern, a value sequence, a categorical column gets a frequency chart, and its relation to each target of the analysis, and to the segment column when there is one, gets the chart that suits the pair, with the target on the vertical axis where it is a number. A column is not charted against every other column; which columns move together is shown under Strongest Correlations. The charts complement the figures in the Description chapter; the shape of a distribution and the position of its outliers are easier to see than to read.
5.3 In2
In2: 0 to 9, median 5, 10 distinct values; moves with In1 most (r 0.763).
5.3.1 Description
| Statistic | Value |
|---|---|
| Type | integer |
| Values present | 488 |
| Distinct values | 10 |
| Continuous | yes |
| Decimal places | 0 |
| Minimum | 0 |
| Maximum | 9 |
| Mean | 5.084 |
| Median | 5 |
| Mode | 6 |
The first 10 rows of 17. The other 7 are in the notebook.
5.3.1.1 By Value
In2 accounts for 55% of the spread in Target: 8 averages 4.84 above 1 (95% interval 4.21 to 5.47). An association, not a cause.
| Target | Value | Rows | Average or share | 95% interval |
|---|---|---|---|---|
| Target | 8 | 25 | 7.04 | 6.587 to 7.493 |
| Target | 7 | 57 | 6.421 | 6.151 to 6.691 |
| Target | 6 | 121 | 5.95 | 5.774 to 6.127 |
| Target | 5 | 115 | 5.304 | 5.124 to 5.484 |
| Target | 4 | 88 | 4.443 | 4.254 to 4.633 |
| Target | 3 | 58 | 3.69 | 3.491 to 3.888 |
| Target | 2 | 16 | 2.938 | 2.371 to 3.504 |
| Target | 1 | 5 | 2.2 | 1.645 to 2.755 |
| In2 | Rows | In1 | In4 | In3 | Target |
|---|---|---|---|---|---|
| 0 | 1 | 5 | 2 | 4 | |
| 1 | 5 | 2.2 | 4.2 | 3.4 | 2.2 |
| 2 | 16 | 3 | 5.063 | 3.625 | 2.938 |
| 3 | 58 | 3.948 | 5.017 | 4.241 | 3.69 |
| 4 | 88 | 4.648 | 5.386 | 4.83 | 4.443 |
| 5 | 115 | 5.626 | 5.609 | 5.391 | 5.304 |
| 6 | 121 | 6.165 | 5.884 | 5.818 | 5.95 |
| 7 | 57 | 6.877 | 6.035 | 6.509 | 6.421 |
| 8 | 25 | 7.6 | 6.16 | 6.16 | 7.04 |
| 9 | 2 | 7.5 | 5.5 | 5.5 | 6.5 |
5.3.1.2 Correlation
| Column | Pair of | Association | Pearson's r | 95% CI (Pearson) | Spearman's ρ | Kendall's τ | q-value | Rows |
|---|---|---|---|---|---|---|---|---|
| In1 | two numbers | 0.7628 | 0.7628 | 0.723 to 0.7976 | 0.7492 | 0.6452 | < 0.001 | 488 |
| Target | two numbers | 0.74 | 0.739 | 0.6959 to 0.7768 | 0.74 | 0.6368 | < 0.001 | 488 |
| In3 | two numbers | 0.6573 | 0.6573 | 0.6038 to 0.7049 | 0.6558 | 0.5582 | < 0.001 | 488 |
| In4 | two numbers | 0.38 | 0.3784 | 0.2997 to 0.452 | 0.38 | 0.3164 | < 0.001 | 488 |
5.3.2 Visualization
The rows of In2 are not charted in their order: consecutive rows are no more alike than rows picked at random (a correlation of -0.06 between each value and the next), so the order carries no pattern to show.
5.4 In3
In3: 2 to 8, median 5, skewed to high values, 7 distinct values; moves with Target most (r 0.823).
5.4.1 Description
| Statistic | Value |
|---|---|
| Type | integer |
| Values present | 488 |
| Distinct values | 7 |
| Continuous | yes |
| Decimal places | 0 |
| Minimum | 2 |
| Maximum | 8 |
| Mean | 5.344 |
| Median | 5 |
| Mode | 6 |
The first 10 rows of 17. The other 7 are in the notebook.
5.4.1.1 By Value
In3 accounts for 68% of the spread in Target: 8 averages 5.67 above 2 (95% interval 4.30 to 7.03). An association, not a cause.
| Target | Value | Rows | Average or share | 95% interval |
|---|---|---|---|---|
| Target | 8 | 4 | 8 | 6.701 to 9.299 |
| Target | 7 | 62 | 6.887 | 6.657 to 7.117 |
| Target | 6 | 174 | 5.977 | 5.856 to 6.098 |
| Target | 5 | 143 | 4.811 | 4.674 to 4.948 |
| Target | 4 | 74 | 3.784 | 3.634 to 3.934 |
| Target | 3 | 25 | 2.84 | 2.532 to 3.148 |
| Target | 2 | 6 | 2.333 | 1.249 to 3.417 |
| In3 | Rows | In2 | In1 | In4 | Target |
|---|---|---|---|---|---|
| 2 | 6 | 1.667 | 2.333 | 4.333 | 2.333 |
| 3 | 25 | 3.16 | 3.56 | 4.28 | 2.84 |
| 4 | 74 | 3.811 | 4.311 | 4.838 | 3.784 |
| 5 | 143 | 4.755 | 5.203 | 5.462 | 4.811 |
| 6 | 174 | 5.747 | 6.103 | 5.994 | 5.977 |
| 7 | 62 | 6.484 | 6.887 | 6.371 | 6.887 |
| 8 | 4 | 7 | 8 | 7 | 8 |
5.4.1.2 Correlation
| Column | Pair of | Association | Pearson's r | 95% CI (Pearson) | Spearman's ρ | Kendall's τ | q-value | Rows |
|---|---|---|---|---|---|---|---|---|
| Target | two numbers | 0.8268 | 0.8234 | 0.7925 to 0.85 | 0.8268 | 0.7457 | < 0.001 | 488 |
| In2 | two numbers | 0.6573 | 0.6573 | 0.6038 to 0.7049 | 0.6558 | 0.5582 | < 0.001 | 488 |
| In1 | two numbers | 0.6493 | 0.6493 | 0.5948 to 0.6978 | 0.6343 | 0.5388 | < 0.001 | 488 |
| In4 | two numbers | 0.6256 | 0.6256 | 0.5684 to 0.6768 | 0.6229 | 0.5477 | < 0.001 | 488 |
5.4.2 Visualization
The rows of In3 are not charted in their order: consecutive rows are no more alike than rows picked at random (a correlation of -0.06 between each value and the next), so the order carries no pattern to show.
5.5 In4
In4: 2 to 8, median 6, skewed to low values, 7 distinct values; moves with Target most (r 0.751).
5.5.1 Description
| Statistic | Value |
|---|---|
| Type | integer |
| Values present | 488 |
| Distinct values | 7 |
| Continuous | yes |
| Decimal places | 0 |
| Minimum | 2 |
| Maximum | 8 |
| Mean | 5.611 |
| Median | 6 |
| Mode | 6 |
The first 10 rows of 17. The other 7 are in the notebook.
5.5.1.1 By Value
In4 accounts for 55% of the spread in Target: 8 averages 5.17 above 3 (95% interval 3.06 to 7.28). An association, not a cause.
| Target | Value | Rows | Average or share | 95% interval |
|---|---|---|---|---|
| Target | 8 | 4 | 8 | 5.75 to 10.25 |
| Target | 7 | 72 | 6.875 | 6.674 to 7.076 |
| Target | 6 | 202 | 5.673 | 5.536 to 5.81 |
| Target | 5 | 160 | 4.475 | 4.329 to 4.621 |
| Target | 4 | 42 | 3.405 | 3.147 to 3.663 |
| Target | 3 | 6 | 2.833 | 2.043 to 3.623 |
| In4 | Rows | In2 | In1 | In3 | Target |
|---|---|---|---|---|---|
| 2 | 2 | 3.5 | 5 | 2.5 | 1.5 |
| 3 | 6 | 3.833 | 3.667 | 3.5 | 2.833 |
| 4 | 42 | 4.095 | 4.238 | 4.167 | 3.405 |
| 5 | 160 | 4.619 | 5.013 | 4.794 | 4.475 |
| 6 | 202 | 5.366 | 5.832 | 5.733 | 5.673 |
| 7 | 72 | 5.972 | 6.5 | 6.319 | 6.875 |
| 8 | 4 | 6.5 | 7.25 | 6.75 | 8 |
5.5.1.2 Correlation
| Column | Pair of | Association | Pearson's r | 95% CI (Pearson) | Spearman's ρ | Kendall's τ | q-value | Rows |
|---|---|---|---|---|---|---|---|---|
| Target | two numbers | 0.752 | 0.7512 | 0.7098 to 0.7875 | 0.752 | 0.672 | < 0.001 | 488 |
| In3 | two numbers | 0.6256 | 0.6256 | 0.5684 to 0.6768 | 0.6229 | 0.5477 | < 0.001 | 488 |
| In1 | two numbers | 0.4438 | 0.4423 | 0.368 to 0.511 | 0.4438 | 0.3743 | < 0.001 | 488 |
| In2 | two numbers | 0.38 | 0.3784 | 0.2997 to 0.452 | 0.38 | 0.3164 | < 0.001 | 488 |
5.5.2 Visualization
The rows of In4 are not charted in their order: consecutive rows are no more alike than rows picked at random (a correlation of -0.01 between each value and the next), so the order carries no pattern to show.
5.6 Target
Target: 1 to 9, median 5, 9 distinct values; moves with In3 most (r 0.823).
5.6.1 Description
| Statistic | Value |
|---|---|
| Type | integer |
| Values present | 488 |
| Distinct values | 9 |
| Continuous | yes |
| Decimal places | 0 |
| Minimum | 1 |
| Maximum | 9 |
| Mean | 5.23 |
| Median | 5 |
| Mode | 6 |
The first 10 rows of 17. The other 7 are in the notebook.
5.6.1.1 By Value
| Target | Rows | In2 | In1 | In4 | In3 |
|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 4 | 2.5 |
| 2 | 12 | 2 | 2.417 | 3.833 | 3 |
| 3 | 38 | 3.237 | 3.921 | 4.526 | 3.763 |
| 4 | 100 | 4.01 | 4.43 | 5 | 4.46 |
| 5 | 116 | 4.974 | 5.293 | 5.474 | 5.284 |
| 6 | 135 | 5.807 | 6.148 | 6.015 | 6.007 |
| 7 | 62 | 6.516 | 7.016 | 6.565 | 6.387 |
| 8 | 19 | 7.053 | 7.737 | 6.737 | 6.789 |
| 9 | 4 | 7 | 8 | 7.5 | 7.25 |
5.6.1.2 Correlation
| Column | Pair of | Association | Pearson's r | 95% CI (Pearson) | Spearman's ρ | Kendall's τ | q-value | Rows |
|---|---|---|---|---|---|---|---|---|
| In3 | two numbers | 0.8268 | 0.8234 | 0.7925 to 0.85 | 0.8268 | 0.7457 | < 0.001 | 488 |
| In1 | two numbers | 0.7531 | 0.7499 | 0.7083 to 0.7864 | 0.7531 | 0.655 | < 0.001 | 488 |
| In4 | two numbers | 0.752 | 0.7512 | 0.7098 to 0.7875 | 0.752 | 0.672 | < 0.001 | 488 |
| In2 | two numbers | 0.74 | 0.739 | 0.6959 to 0.7768 | 0.74 | 0.6368 | < 0.001 | 488 |
5.6.2 Visualization
The rows of Target are not charted in their order: consecutive rows are no more alike than rows picked at random (a correlation of -0.06 between each value and the next), so the order carries no pattern to show.
5.7 Duplicate Rows
Across 488 rows the table has 317 exact-duplicate rows (65% of the table; biggest duplicate group 11 rows): substantial. Likely root causes: repeated CSV imports, a missing primary-key deduplication step, or row-level leakage between intended-distinct sources. Deduplicating before training matters; duplicated rows inflate model confidence artificially.
- Largest group size
- 11
- N exact duplicates
- 317
- N exact groups
- 92
- N rows
- 488
- N unique
- 263
- Pct exact duplicates
- 64.96
- Verdict
- Substantial
How to read this
This chapter counts rows that repeat another row exactly, value for value: how many rows take part in a duplicate group, how many groups exist, and how large the largest group is.
Duplicate rows inflate the apparent row count and can place identical rows on both sides of the training and held-out split, which flatters every quality figure in this report. Only exact repeats are detected here; rows that differ by a rounding digit or a whitespace variant count as distinct.
5.8 Leakage Suspicion
Nothing to check. No column that awaits an answer tracks a target at 0.95 or more, so no column on its own gives a target away.
- Critical threshold
- 0.999
- N confirmed
- 0
- N critical
- 0
- N suspects
- 0
- Threshold
- 0.95
- Verdict
- Clean
How to read this
For every column that is not a target, this chapter reads how closely the column tracks each target, on an association from 0 to 1, and lists any pair at 0.95 or more. For two numeric columns the association is the stronger of Pearson's r and Spearman's rho; when the column or the target is categorical it is the correlation ratio η, or Cramér's V when both are. At 0.999 or more the column reproduces the target: that is what a copy or a rescaled copy of it looks like.
A listed column poses one question: is its value known before the target's value exists? If it is, the column is a strong input and the quality stated in this report is real. If it was computed from the target or recorded after it, a formula that reads it looks right on these rows and will not hold on new data. The coefficients read the same in both cases; the answer needs someone who knows where the column came from.
The owner of the analysis can give that answer. A column confirmed as known before the target is listed as confirmed and is no longer a suspect. A column that turns out to be a leak is left out of the model's inputs, and the analysis run again.
5.9 General
Clustering found no clear groups. The best grouping of the 488 complete rows over 4 numeric columns is into 2 groups, with a silhouette of 0.371 (on a scale where 1 is fully separate groups and 0 is none), below the 0.50 that clear structure shows, so no Clusters chapter is written.
| Groups | Silhouette | Inertia | Kept |
|---|---|---|---|
| 2 | 0.3709 | 1058.3 | yes |
| 3 | 0.2672 | 840.3 | |
| 4 | 0.278 | 706.9 | |
| 5 | 0.2385 | 633.3 | |
| 6 | 0.2412 | 572.3 |
- Inertia
- 1058.3
- K
- 2
- N columns used
- 4
- N rows
- 488
- N rows complete
- 488
- Silhouette
- 0.3709
- Verdict
- Weak
How to read this
The General chapter covers what holds across all columns together: descriptive statistics, correlations, missing values and outlier ranges under Description, the Clusters found among the rows, and, when a segment column was chosen, the Segments.
5.9.1 Description
| Variable | Type | Values present | Distinct values | Continuous | Decimal places |
|---|---|---|---|---|---|
| In2 | integer | 488 | 10 | yes | 0 |
| In1 | integer | 488 | 10 | yes | 0 |
| In4 | integer | 488 | 7 | yes | 0 |
| In3 | integer | 488 | 7 | yes | 0 |
| Target | integer | 488 | 9 | yes | 0 |
| Variable | Mean | Median | Mode |
|---|---|---|---|
| In2 | 5.084 | 5 | 6 |
| In1 | 5.506 | 6 | 6 |
| In4 | 5.611 | 6 | 6 |
| In3 | 5.344 | 5 | 6 |
| Target | 5.23 | 5 | 6 |
| Variable | Minimum | Maximum | Standard deviation | Variance | Interquartile range | Sum |
|---|---|---|---|---|---|---|
| In2 | 0 | 9 | 1.551 | 2.406 | 2 | 2481 |
| In1 | 0 | 9 | 1.526 | 2.329 | 1.25 | 2687 |
| In4 | 2 | 8 | 0.9375 | 0.8789 | 1 | 2738 |
| In3 | 2 | 8 | 1.136 | 1.29 | 1 | 2608 |
| Target | 1 | 9 | 1.422 | 2.021 | 2 | 2552 |
| Variable | Skewness | Kurtosis | Total sum of squares |
|---|---|---|---|
| In2 | -0.1904 | -0.1976 | 1171.6 |
| In1 | -0.1544 | 0.0268 | 1134 |
| In4 | -0.3835 | 0.5521 | 428 |
| In3 | -0.4598 | 0.03806 | 628.2 |
| Target | -0.1002 | -0.119 | 984.3 |
How to read this
This chapter summarizes every column at once: the descriptive statistics table below, then Correlation, Missing Values and Outliers. For a numeric column the table reports its center (mean, median, mode), its spread (standard deviation, variance, interquartile range, percentiles) and its shape (skewness, kurtosis); for a categorical column it reports the count and share of each category.
Each row of the tables describes one column. The statistics are grouped: the column itself, where its values center, how far they spread, what shape their distribution takes, and its percentiles. They read as follows.
| Statistic | Explanation |
|---|---|
| Variable | Name of the column the statistics in the row describe |
| Type | Type assigned during data processing (bool, integer, number, decimal, text or date) |
| Values present | Number of rows in which the column holds a value |
| Distinct values | Number of different values the column holds |
| Continuous | Whether the column can take any value in a range, as opposed to a discrete column limited to a fixed set of values |
| Decimal places | Length of the fractional part of the values |
| Mean | Arithmetic average: the sum of all values divided by the number of values |
| Median | The middle value: half of the values lie at or below it and half at or above it. It is also the 50% percentile |
| Mode | The value that appears most often |
| Minimum | Smallest numeric value in the column |
The first 10 rows of 20. The other 10 are in the notebook.
5.9.1.1 Missing Values
No column has a missing value: all 488 rows are complete.
- N columns with missing
- 0
- N complete rows
- 488
- N rows
- 488
- Pct complete rows
- 100
How to read this
The table lists, for each column, the count of present values (Values) beside the row count of the processed table (Rows), the number of missing values and their share. The larger the share, the less any statistic or model built on the column can be trusted.
- Below 5% missing: safe to use as it is.
- 5% to 25% missing: usable; filling the gaps or excluding those rows is worth considering.
- 25% or more missing: the column is flagged and receives its own chapter in the Data section. Verify why so much is missing before trusting any result that depends on it.
A model is computed only on rows that hold a value in every column it reads, so gaps spread over several columns cost more rows than any one column's share suggests. The last column, Complete rows without, shows for each input with gaps how many rows would be complete if that column were left out; it is blank for a column without gaps, for a target, and for a column no model reads. A column that is the only gap in many rows is the one whose removal keeps the most data.
5.9.1.2 Outliers
4 of 5 numeric columns have values beyond the Tukey fences (1.5 times the interquartile range outside the quartiles). in1 has the largest share: 14% of its values. Whether such values are rare but real rows or recording errors decides whether they are kept.
| Column | Values | Lower fence | Upper fence | Below | Above | Share |
|---|---|---|---|---|---|---|
| In1 | 488 | 2.875 | 7.875 | 15 | 55 | 14.34% |
| In2 | 488 | 1 | none above | 1 | 0 | 0.2% |
| In3 | 488 | 3.500 | 7.500 | 31 | 4 | 7.17% |
| In4 | 488 | 3.500 | 7.500 | 8 | 4 | 2.46% |
| Target | 488 | none below | none above | 0 | 0 | 0% |
- N columns
- 5
- N columns assessed
- 5
- N columns with outliers
- 4
- Worst column
- in1
- Worst pct
- 14.34
How to read this
The table counts the outliers of every numeric column by the Tukey rule. Q1 and Q3 are the values below which 25% and 75% of a column's values fall, and the interquartile range is Q3 minus Q1. The lower fence lies 1.5 times the interquartile range below Q1, the upper fence the same distance above Q3; a value beyond a fence is an outlier. For each column the table gives both fences, how many values lie below and above them, their share of the column's values, and the single value that lies furthest beyond its fence. These are the same fences the box plots draw their whiskers to.
A column is not assessed when it has fewer than four values, or when the middle half of its values sits on a single number: the interquartile range is then zero and the rule would call every other value an outlier.
Extreme values can be real (rare but legitimate rows) or errors (measurement mistakes, data-entry slips). Which case applies decides whether to keep, clip or drop them before trusting a model that uses the column. A strongly skewed column has many values beyond one fence without any of them being wrong.
5.9.1.3 Correlation
10 column pairs analyzed across 488 rows; 10 (100%) are unlikely to be chance, after allowing for how many pairs were tested. Any other pair may still show a large association, but one that cannot be told apart from chance on the rows the pair was measured on. An association states that two columns go together, not that one brings the other about.
- N categorical pairs
- 0
- N mixed pairs
- 0
- N numeric pairs
- 10
- N pairs
- 10
- N samples
- 488
- N significant
- 10
| Association | 95% CI | Column | Kendall | Kind | N | Other column | Pearson | Q value | Spearman |
|---|---|---|---|---|---|---|---|---|---|
| 0.8268 | 0.7925, 0.85 | In3 | 0.7457 | numeric | 488 | Target | 0.8234 | < 0.001 | 0.8268 |
| 0.7628 | 0.723, 0.7976 | In1 | 0.6452 | numeric | 488 | In2 | 0.7628 | < 0.001 | 0.7492 |
| 0.7531 | 0.7083, 0.7864 | In1 | 0.655 | numeric | 488 | Target | 0.7499 | < 0.001 | 0.7531 |
| 0.752 | 0.7098, 0.7875 | In4 | 0.672 | numeric | 488 | Target | 0.7512 | < 0.001 | 0.752 |
| 0.74 | 0.6959, 0.7768 | In2 | 0.6368 | numeric | 488 | Target | 0.739 | < 0.001 | 0.74 |
| 0.6573 | 0.6038, 0.7049 | In2 | 0.5582 | numeric | 488 | In3 | 0.6573 | < 0.001 | 0.6558 |
| 0.6493 | 0.5948, 0.6978 | In1 | 0.5388 | numeric | 488 | In3 | 0.6493 | < 0.001 | 0.6343 |
| 0.6256 | 0.5684, 0.6768 | In3 | 0.5477 | numeric | 488 | In4 | 0.6256 | < 0.001 | 0.6229 |
| 0.4438 | 0.368, 0.511 | In1 | 0.3743 | numeric | 488 | In4 | 0.4423 | < 0.001 | 0.4438 |
| 0.38 | 0.2997, 0.452 | In2 | 0.3164 | numeric | 488 | In4 | 0.3784 | < 0.001 | 0.38 |
How to read this
This chapter measures how pairs of columns go together: the Correlation Measures table below, the Strongest Correlations, the Correlation Heatmap and, when the analysis has targets, the Potential Data Leaks. Numeric columns and categorical columns are both paired, a categorical column being one that is not numeric and holds between 2 and 50 distinct values. Every pair is stated as an association from 0 (unrelated) to 1 (one column gives the other away), so pairs of different kinds can be compared. An association states that two columns go together, not that one brings the other about.
The table lists every pair of columns once, with its kind, its association, the figures of its kind, its q-value and the number of rows the pair was measured on. A pair is measured on the rows where both columns hold a value.
| Measure | Explanation |
|---|---|
| Kind | What the two columns hold: two numbers (numeric), a number and a category (numeric_categorical), or two categories (categorical) |
| Association | Strength of the pair from 0 to 1, whatever its kind. Two numbers: the stronger of Pearson's coefficient and Spearman's ρ, without its sign. A number and a category: the correlation ratio η. Two categories: Cramér's V |
| Covariance | Two numbers only. Direction of the relationship (positive or negative) in the units of the two columns. It is not standardized, so its size says little about strength without further context |
| Pearson's Correlation Coefficient | Two numbers only. Strength and direction of a linear relationship, from -1 (perfect negative) through 0 (no linear relationship) to 1 (perfect positive) |
| Spearman's ρ | Two numbers only. Strength and direction of a monotonic relationship, computed on ranks rather than values. It captures relationships that consistently rise or fall even when they are not linear, and it is less affected by extreme values |
| Kendall's τ | Two numbers only. Strength and direction of a monotonic relationship based on the concordance of ranks between the two columns. More conservative than Spearman's ρ, and steadier on small tables |
| Correlation ratio η | A number and a category. The square root of the share of the number's spread that lies between the categories, from 0 (every category has the same average) to 1 (the category fixes the number). For a category of two values it equals the absolute Pearson coefficient with the category coded 0 and 1. Its square, η², is that share: an η of 0.6 means the categories account for 36% of the spread |
| Cramér's V | Two categories. How far the table of counts of the two columns departs from what unrelated columns would give, from 0 (unrelated) to 1 (each value of one column goes with one value of the other), corrected so that columns with many rare values do not read as related |
| q-value | How likely a pair this strong is among unrelated columns, adjusted for the number of pairs listed (Benjamini-Hochberg false-discovery rate). Among all pairs reported with q below 0.05, about 5% or fewer are expected to be false discoveries. The right measure for ranking many pairs |
| 95% CI (Pearson) | Two numbers only. Lower and upper bounds of the 95% confidence interval on the Pearson coefficient, via the Fisher z-transform, on the rows the pair was measured on. A narrow interval means the coefficient is well estimated; a wide interval that straddles zero means the relationship is uncertain |
The first 10 rows of 12. The other 2 are in the notebook.
5.9.1.3.1 Strongest Correlations
The 20 pairs of columns with the smallest q-values: the associations that remain significant after the Benjamini-Hochberg adjustment for the number of pairs tested. Among all pairs reported with q below 0.05, about 5% or fewer are expected to be false discoveries; a pair above 0.05 may be coincidence even when its association looks strong.
A pair with an association above 0.9 often signals a duplicate or derived column. Moderately strong pairs with a low q-value are the relationships most worth investigating.
Each listed pair is drawn below, the picture behind the figure. Two numbers are drawn as a scatter, one point per row, the quickest way to see whether a strong value rests on a trend or on a few rows far from the rest. A number and a category are drawn as the number's box plot per category. Two categories are drawn as a heat map of how many rows hold each combination of values.
5.9.1.3.3 Potential Data Leaks
The table lists the association between every column and each target, whatever the two hold, with its q-value. An association close to 1 with a target, together with a q-value near zero, is one of two things: a strong input, or a data leak, a column derived from the target after the fact, such as a tax amount computed from the price being predicted. A leaked column trains a model that looks perfect and fails on new data, because the leak is not available at prediction time.
The Leakage Suspicion chapter in the Data section lists the pairs at 0.95 or more and marks those at 0.999 or more, where a column reproduces the target. For two numbers the association is whichever of Pearson's r and Spearman's rho is the stronger, because a column derived from the target often follows a curve rather than a line: the log of a price tracks the price perfectly, and Pearson alone reads that as weaker than it is. For a number and a category it is the correlation ratio η, and for two categories Cramér's V. Whether a listed column is a strong input or a leak depends on whether it is known before the target exists, which the owner of the analysis can confirm.
5.10 Findings
1 finding from the data profile (1 Critical) is listed in the Concerns chapter at the top of the analysis.
6 Implications
The Implications section states how to act on the models: the Stakeholder Summary and the Recommended Next Data, where the models hold and where they weaken (Quality by Input Range, Rows unlike the ones the model learned from, Per-Segment Predictions, Fairness), what to weigh before relying on them (Causal Honesty, Deployment Readiness), and the Provenance & Audit trail a reviewer needs to retrace the analysis.
6.1 Stakeholder Summary
The model analyzes Target using three inputs and finds that the strongest input is In4, which makes up 71% of the model's reliance. The model explains 85% of the variation in Target, and moving In4 from 5.00 to 6.00 raises the model's output by about 0.45 on average, with everything else as observed. Compared with a Poisson regression on the numeric inputs, the model removes 16 fewer points of Poisson deviance and the two are about level, with the model ahead on 49% of the held-out rows.
The model did not beat the baseline Poisson regression, so its findings are not an advance. The check rests on 98 held-out rows and 3.1% lay outside the range the model was trained on. The held-out split is likely contaminated by the 65% exact duplicate rows in the data. Validate on a fresh sample before relying on the model for decisions.
6.2 Recommended Next Data
- Collect new, non-duplicate rows for Target. Reason: the data has 65% exact duplicate rows, so the held-out split is likely contaminated, which limits what the model's quality figure says about generalization.
- Inspect why the model does not beat a Poisson regression on the numeric inputs for Target. Reason: the model does not beat a Poisson regression on the numeric inputs on Target, and the lift of -16 points (Poisson deviance removed) is negligible.
6.3 Quality by Input Range
Quality by Input Range splits the held-out rows by every usable input and compares the model's quality on each segment with its quality on all held-out rows. For a number the quality is R², the share of variation explained; for a yes-or-no target it is accuracy, the share of rows classified correctly, the measure the Classification Performance chapter reports. A numeric input is split into quartiles; a categorical input with at most ten values is split by value. A segment with fewer than three held-out rows is left out, because its gap would be noise.
The chart is sorted from the worst gap to the best. Each bar is the segment's figure minus the figure on all held-out rows.
- Negative bars: the model is weaker on these rows than on all held-out rows. A segment 0.25 or more below the overall value (25 points of accuracy) reads much weaker; a segment 0.10 to 0.25 below reads weaker. These are the segments to treat with caution.
- Positive bars: the model is stronger on these rows than overall. A strong segment can mask weakness elsewhere, so the negative segments come first.
6.3.1 Target by Input Range
Segment in2 in the third quarter (5) performs in line with the aggregate (20 held-out rows): segment R² 0.667 against aggregate R² 0.848 (gap -0.181). No quality drift detected.
| Segment | Rows | R² | Gap to all rows | Verdict |
|---|---|---|---|---|
| In2 = q3 | 20 | 0.6671 | -0.1812 | In line with all rows |
| In3 = q4 | 47 | 0.6355 | -0.2128 | In line with all rows |
| In1 = q4 | 59 | 0.7204 | -0.1279 | In line with all rows |
| In4 = q4 | 52 | 0.6593 | -0.1889 | In line with all rows |
| In3 = q1 | 25 | 0.4853 | -0.3630 | In line with all rows |
| In3 = q3 | 26 | 0.5228 | -0.3255 | In line with all rows |
| In2 = q4 | 49 | 0.7392 | -0.1091 | In line with all rows |
| In4 = q2 | 33 | 0.7665 | -0.0817 | In line with all rows |
| In1 = q1 | 25 | 0.7738 | -0.0744 | In line with all rows |
| In2 = q2 | 14 | 0.6528 | -0.1955 | Too few rows to judge |
The first 10 rows of 13. The other 3 are in the notebook.
- Aggregate MAE
- 0.4284
- Aggregate R²
- 0.8483
- Measure
- r2
- Target
- Target
6.4 Rows unlike the ones the model learned from
3 of 98 held-out rows (3.1%) lie further from the training rows than 99 in 100 rows like them would, measured on the 3 inputs the model reads (in2, in3, in4). That is a small minority; the held-out figures describe the model on rows like the ones it was trained on.
| Target | Inputs measured | Line | Held-out rows | Beyond the line | Share |
|---|---|---|---|---|---|
| Target | In2, In3, In4 | 3.368 | 98 | 3 | 3.06% |
| Count | From | To |
|---|---|---|
| 0 | 0 | 0.135 |
| 0 | 0.135 | 0.269 |
| 0 | 0.269 | 0.404 |
| 4 | 0.404 | 0.539 |
| 0 | 0.539 | 0.674 |
| 8 | 0.674 | 0.808 |
| 13 | 0.808 | 0.943 |
| 13 | 0.943 | 1.078 |
| 0 | 1.078 | 1.213 |
| 0 | 1.213 | 1.347 |
The first 10 rows of 24. The other 14 are in the notebook.
How to read this
Out-of-distribution detection finds held-out rows that lie far from the training rows of a model. It looks only at the numeric inputs the model reads. Each of them is standardized against the mean and standard deviation of the training rows, and the standardized values are combined into one distance per row. The line a row is read against depends on the number of inputs: it is the distance that 99 in 100 rows like the training rows stay below. A row beyond the line is flagged as out of distribution, and the held-out figures are unlikely to apply to it. When the held-out rows resemble the training rows, about one row in a hundred is flagged.
The chart shows one bar per held-out row in row order; the flagged rows are the bars at or above the line. With more than one target, the chart shows the model with the largest flagged share, and the figures name it.
The check treats inputs independently, so it will not notice a rare combination of values that are each ordinary on their own, such as a large height with a small weight. It is a first screen. A decision that depends on catching unusual inputs needs a dedicated check.
6.5 Causal Honesty
Diafunc found a formula that captures the structure in the training data. That structure is correlative: it states what predicts what in this data. It does not establish that any input causes the target.
A correlative model is the right tool for predicting the target on new rows like these, for exploring which inputs carry information about the target, and for an early-stage diagnostic or screening tool.
A correlative model is not the right tool for claiming that intervening on an input will change the target, for attributing a target's values to specific causes for accountability, or for choosing between policies that act on the inputs.
The Quality by Input Range, Fairness and Concerns chapters show the most common confounding patterns in the data.
The model is correlative: it captures which columns predict the target in the training data, not whether changing one would change the other. Predictions are reliable for inputs that look like the training distribution; they are not evidence that intervening on an input will change the target. The 4 upgrade pathways below show what evidence would support a causal claim.
| Name | Summary |
|---|---|
| Controlled experiment | Randomize the input to be acted on, then analyze again. Randomization breaks confounding, so any change in the target is attributable to the input. |
| Natural experiment | Find a setting where the input varied for reasons unrelated to the target, an instrumental variable. Variation of that kind supports causal inference. |
| Causal graph | Write out what plausibly causes what, then check whether the model's terms are consistent with the conditional independences the graph implies. |
| Control for confounders | If a third variable plausibly moves both the input and the target, stratify by it or condition on it before training again. Gaps in the Quality by Input Range chapter are a clue. |
6.6 Deployment Readiness
The analysis ran with the configuration below. The checks and the verdict of this chapter are read against it.
- Targets: 1 target:
target - Rows: 488 rows
- Per-Segment Predictions: not produced, because no segment column was chosen
- Fairness: not assessed, because the data has no categorical column other than the target
- Out-of-Distribution: held-out rows are scored against the training rows; predictions made outside this report are not scored
- Provenance & Audit: recorded in the chapter of that name
Across 4 checks (3 pass, 1 warns, 0 fail, 1 critical concern), every check passes or warns, but a critical concern is open: Not ready to use yet. The Concerns chapter names what is open; it needs addressing before the predictions are used.
| Name | Status | Detail |
|---|---|---|
| Generalization | Pass | held-out quality holds up against training quality |
| Baseline Lift for Target | Warn | is not told apart from the best yardstick by 16% of its Poisson deviance |
| Model Constraints | Pass | every held-out prediction is finite and within the training range |
| Out-of-Distribution | Pass | 3.1% of held-out rows are out of distribution |
- N critical concerns
- 1
- N fail
- 0
- N pass
- 3
- N warn
- 1
- Overall verdict
- Not ready
6.7 Provenance & Audit
This record captures the inputs and configuration that produced the report, so the analysis can be reproduced or the result defended in an audit.
| Item | Value |
|---|---|
| Analysis ID | 764da0d0-48b7-11f1-8087-d378640fe787 |
| Run time | 2026-10-28 14:48 UTC |
| Total rows | 488 |
| Modeled targets | Target |
| Segment column | not available |
| Time limit | 1 h 0 min |
| Time taken | 44 min 15 s |
| Credit budget | 500 credits |
| Ufinq version | 0.3.48 |