Model Fitting and Uncertainty

Purpose¶

Model fitting turns a scientific question into a comparison between data and a forward model. In COMP 521, that comparison must include diagnostics and an uncertainty statement, not only a best-fit line or parameter value.

Planned sequence¶

1. build and inspect a design matrix;

2. fit a model while checking conditioning and residuals;

3. quantify uncertainty with resampling and model diagnostics;

4. state what the data support, and what they do not.

The foundations are linear algebra, probability, and statistical tools.