Measuring Economic Growth from Space: Predicting GDP Discrepancies with Nighttime Lights

Walk through the full pipeline (data → split → CV → 5 models → compare) on a small simulated dataset where we know the true DGP (5informative features + 25 noise). Easier to see what each line does — and what each method picks. Replication of Muchlinski, Siroky, He & Kocher 2016 (Political Analysis), "Comparing Random Forests with Logistic Regression for Predicting Class-Imbalanced Civil War Onset Data