Guides

A real case from the table to the equation: 308 measurements, six inputs, one readable model, and what the report says about it.

What the method is, how a search for a formula works, where it helps and where it does not, and what it cannot promise.

Which rows are held out, what the figures of a model say, what it is measured against, and the checks a report runs on its own model.

What exists, each tool in its own description, the kinds of tool beside it, and how to choose, including when the answer is not us.

Connect Claude or another assistant to Diafunc over MCP, start an analysis in one sentence, and read the report and the model back.

What a neural network returns and what a symbolic model returns, where each is at home, and how to choose.