9.9 Summary

This chapter introduced the principles and workflow of model building in BI. We distinguished three types of models — descriptive (what happened), predictive (what will happen), and prescriptive (what to do) — and traced the modeling workflow from problem definition through deployment and monitoring.

The core tension in model building is between fit and generalizability: a model must be complex enough to capture real patterns but simple enough to avoid overfitting to noise. We demonstrated this trade-off with R code, showing how a simple model can outperform a complex one on unseen data. We also covered validation techniques — from R² and AIC for in-sample assessment to holdout testing and cross-validation for out-of-sample evaluation.

In the next chapter, we apply these principles hands-on, building linear and logistic regression models on the Absenteeism at Work dataset.