Dataset and Code for Forecasting Illicit Practice Impacts on Economies of Conflict Countries: XGBoost Approach

This repository contains the dataset and Python code used for forecasting Illicit practice impacts on economies of conflict countries using XGBoost approach. The study applies XGBoost regression models to estimate the relationship between: CPI (Corruption Perceptions Index); GML (Global Money Laundering Index); SIS (Cybersecurity-related indicator); GPI (Global Peace Index – conflict proxy); the dependent variable: Logarithmic GDP growt. The repository includes: Data preprocessing procedures Variable transformations (log growth, percentage growth) Normalization and directional adjustment of indicators Multicollinearity diagnostics (correlation matrix, VIF) Lag selection using AIC and BIC XGBoost model training and evaluation Feature importance analysis SHAP-based interpretability analysis Scenario analysis (+1σ shock to conflict-related variables) Visualization outputs (sparklines, maps, heatmaps) The dataset is structured as a country-year panel covering 2012-2024. This material supports reproducibility of the empirical results presented in the associated research work.