atchirume/factor-augmented-inflation-nowcasting-and-forecasting

Central bank-grade factor-augmented inflation nowcasting and forecasting framework for Zimbabwe. Combines PCA-based macroeconomic factor extraction, machine learning models (Elastic Net, Random Forest, SVR, XGBoost), scenario analysis, and inflation regime classification aligned to SADC targets. # Factor-Augmented Inflation Nowcasting and Forecasting Framework A central bank–grade inflation nowcasting and forecasting system developed by **Chirume Admire Tarisirayi**. This framework integrates **principal component analysis (PCA)**, **econometric modelling**, and **machine learning techniques** to generate robust, forward-looking inflation estimates and policy-relevant insights for Zimbabwe and comparable economies. --- ## 🚀 Overview Macroeconomic environments are characterised by: - High-dimensional data - Structural breaks - Rapidly evolving dynamics Traditional models often fail to capture these complexities. This framework addresses these limitations through: - Factor-based dimensionality reduction - Multi-model machine learning forecasting - Real-time nowcasting capability - Scenario-based policy simulation --- ## ⚙️ Key Features - 📊 **PCA-Based Factor Extraction** Reduces large macroeconomic datasets into latent economic drivers - 📈 **Inflation Nowcasting Engine** Incorporates high-frequency indicators to estimate current-period inflation - 🤖 **Multi-Model Forecasting Framework** Combines: - Elastic Net - Random Forest - Support Vector Regression (SVR) - XGBoost - 🔄 **Scenario Analysis Module** Simulates macroeconomic shocks in: - Exchange rate - Money supply - Fiscal conditions - External price dynamics - 🧠 **Inflation Regime Classification** Aligns inflation dynamics with SADC macroeconomic convergence targets - 📉 **Interactive Dashboard (Streamlit)** Real-time visualisation of forecasts, factor contributions, and scenarios --- ## 🧠 Methodological Framework The model follows a **Factor-Augmented Approach**: \[ \pi_t = \alpha + \beta F_t + \varepsilon_t \] Where: - \( \pi_t \): Inflation - \( F_t \): Extracted macroeconomic factors (via PCA) - \( \beta \): Factor loadings - \( \varepsilon_t \): Error term --- ## 🛠️ Technology Stack - Python - Pandas - Scikit-learn - XGBoost - Plotly - Streamlit --- ## 🌍 Applications This fram …