kampambashula/policy-rate-predictor

Central Bank Policy Rate Forecast Tool is an interactive, Central Bank-grade dashboard designed to Forecast the Bank of Zambia policy rate using machine learning models # Central Bank Policy Rate Forecast Tool Author: Kampamba Shula Year: 2026 ## Overview The Central Bank Policy Rate Forecast Tool is an interactive, Central Bank-grade dashboard designed to: - Forecast the Bank of Zambia policy rate using machine learning models. - Generate Hold / Raise / Lower signals with dynamic, BoZ-targeted commentary. - Explore macroeconomic trends such as inflation, liquidity, lending rates, exchange rates, and broad money (M2). - Support scenario analysis by adjusting macroeconomic inputs. This tool combines time-based ML evaluation with interactive visualization to provide policy insights in a professional, central bank-style format. ## Features Policy Forecasting Predicts the next Bank of Zambia policy rate and generates actionable signals. IMF-Style Commentary Provides detailed insights based on BoZ targets (inflation 6–8%) and key macro indicators. Multiple ML Models Includes Random Forest, Linear Regression, and XGBoost with time-based evaluation. Interactive Scenario Analysis Adjust macroeconomic variables in the sidebar to test alternative policy outcomes. Macro Trends Visualization View historical trends for inflation, liquidity, lending rates, exchange rates, and broad money. Residuals & Model Accuracy Evaluate ML model performance over time, including residuals and actual vs predicted charts. ## Pages in the App Home: Introduction, features, author credits, and instructions. Macro Trends: Interactive charts of historical macroeconomic indicators. Policy Prediction: Forecast the policy rate with IMF-style commentary and key indicators table. ML Model Evaluation: Compare ML model performance (R² scores, residuals, actual vs predicted). ## How It Works Load the preprocessed dataset (final.csv) containing historical macro indicators and BoZ policy rates. Train ML models on an 80:20 time-based split to prevent leakage. Evaluate models and display R² performance, residuals, and prediction accuracy. Generate …