The notebook for this exercise was held back from last week's lab because of the CA2 deadline extension....
An annotated notebook with supporting images, presented in a zip archive (right-click and download as CreditBalance/CreditBalance.zip) has been prepared, based on existing published notebooks but with significant enhancements to the models and also to the discussion and commentary on those models.
The main aspects shown include - encoding categorical features so they can be used in linear regression models - forward selection of features for model building (integrating cross-validation, so that the loss metric is the mean square error on the validation set) - regularisation (ridge and lasso), applied to the overfitted model
Compared to the Diamond Data example from Regression 1, there is much less focus on python and notebooks and more on the actual analysis.
The workbook includes one short exercise:
lambda for ridge regression, analogous to the procedure documented for lasso regression.The data file (right-click and download as FS-CreditBalance/data/Credit.csv) is available for your use.