Skip to content

Linear Regression Visual EDA

🎯 Purpose

This checklist provides a visual workflow for the exploratory data analysis (EDA) phase of a linear regression project. Each step uses a plot to check a key assumption or relationship.

Step 1: Target Variable Distribution

  • [ ] Histogram + KDE
  • [ ] Skewness / Kurtosis calculation

Step 2: Baseline OLS Modeling

  • [ ] Fit model and assess residual plots
  • [ ] QQ plot for normality
  • [ ] Residuals vs Fitted

Step 3: Model Integrity Checks

  • [ ] VIF for multicollinearity
  • [ ] Influence plot / Cook’s Distance
  • [ ] Boxplot of residuals by group

Step 4: Model Extensions

  • [ ] Polynomial fit comparison
  • [ ] Try RLM (Robust Linear Model)
  • [ ] Apply Ridge / Lasso / Elastic Net

Step 5: Model Selection

  • [ ] Plot RMSE vs alpha or complexity
  • [ ] AIC/BIC curve
  • [ ] Residual comparison (train/test)

Step 6: Final Summary & Visual Audit

  • [ ] Screenshot diagnostic visuals

🧠 Final Tip

"Visual EDA for regression is about finding the story in the scatter plots. Look for trends, curves, and clumps before you ever fit a line."