What is multicollinearity and how do you detect and fix it?
StatisticsLevel: FoundationStatistics and probability
What is multicollinearity and how do you detect and fix it?
Hints
- Identify the random variable, sample and estimated parameter.
- State the assumptions; distinguish an estimate from its uncertainty.
Answer
Multicollinearity occurs when two or more predictors in a regression are highly linearly correlated, making it hard to isolate each variable's individual effect — coefficients become unstable and standard errors inflate. Detect it via correlation matrices or Variance Inflation Factor (VIF); VIF > 5–10 is typically flagged as concerning. Fixes: drop/combine correlated features, use PCA to decorrelate, or apply regularization (Ridge/Lasso), which shrinks unstable coefficients.
Source and licence: Ishan Dutta — Awesome Data Science Interview Q&A · MIT