Analytics & Big Data

Session 5: Regression II

Prof. Dr. Gerit Wagner

(2026-04-14)

Understand and assess multiple regression models, including refinement strategies and key assumptions.
Distinguish between inference- and prediction-oriented modeling and their implications for business contexts.
Explain logistic regression as a probabilistic classification model and its role as a bridge to machine learning.

Refining the model

Mulitple regression

example + ceteris paribus

Assumptions

list, illustrate how transformations could model non-linear relationships

Transition: we may accept some violations (e.g., multicollinearity) when we are interested in prediction (statisticians and economists: headache when the goal is to make inferences on individual effects)

Explanation ↔︎ prediction

Correlation vs causality / explanation vs prediction

Example: more garages → higher price?

Regression → Machine learning

illustration

business goal: predict defaults or quick-sales (binary outcome; need to select an appropriate model)

logistic regression…

classification and evaluation metrics (confusion matrix…)

similarity to machine learning

Summary

TODO

Survey: Session 5

https://forms.gle/mVWw3z7ftFn48gDY6