A 5-slide presentation pitching a real-time Next-Word Predictive Text Engine built using the SwiftKey natural language processing corpus for the Johns Hopkins Data Science Capstone.

Tugas Week 11 BINTEL - R Markdown

This report replicates the main results of Hidrobo, Peterman & Heise (2016) — a randomized experiment on cash, food, and voucher transfers and intimate partner violence in Ecuador — and extends the analysis using causal forests (Athey & Wager). We estimate conditional average treatment effects (CATEs) across three covariate specifications motivated by policy realism, validate heterogeneity via 5-fold cross-validation and calibration tests, and run five robustness checks including a placebo test and bootstrap inference.

Final_5_16_26