Armenia-Inflation-Forecasting

Summary

This is a Jupyter Notebook project that compares SARIMA and ARIMAX models for forecasting Armenia's year-over-year inflation rate. It uses macroeconomic variables (exchange rate, policy rate, money supply, oil prices) and structural shock dummies. The analysis includes data cleaning, exploratory data analysis, model selection via grid search, fixed test set evaluation, rolling-origin validation, Diebold-Mariano tests, and scenario analysis. Results indicate the univariate SARIMA model outperformed the multivariate ARIMAX model on the test set.