Summary

In this final chapter on data modeling, we have covered a wide range of topics on regression as a way to model data and make predictions. You learned how to make linear regression models as well as non-linear models, and ways to properly prepare data for such modeling. Metrics such as the Sum of Squared Errors (SSE) and the Root Mean Squared Error (RMSE) were introduced to assess the quality of models fitting data. In addition, visual techniques such as inspecting the histogram of residuals, Q-Q plots, and plotting predicted values versus actual values were shown to be important and easily used tools to determine the quality of a model.

You learned that even with simple linear models, some modest feature engineering such as transforming independent variables (the square root or log, for example) can improve results, at the cost of making it difficult to interpret the model coefficients. The common case of time series data with periodic features (such as daily or weekly)...