Phenoley/electricity-demand-forecasting

XGBoost & LSTM forecasting for West African electricity # electricity-demand-forecasting XGBoost & LSTM forecasting for West African electricity This project develops a machine learning–based forecasting and optimization framework for electricity demand and price dynamics in West African power systems. The study integrates macroeconomic indicators, climate variables, fuel prices, and power system capacity metrics to model real-world energy market behavior. The work is designed to support: Power system planning Tariff analysis Energy policy evaluation Reliability and outage risk assessment 🎯 Objectives Forecast electricity demand (MW) using machine learning models Predict electricity prices (USD/kWh) under varying economic and supply conditions Analyze the impact of capacity constraints, fuel prices, and climate variables Provide a reproducible framework suitable for academic research and policy analysis 🧠 Models Implemented XGBoost Regressor Captures non-linear relationships between demand, price, and explanatory variables LSTM Neural Network Models temporal dependencies in electricity demand and pricing trends 📊 Dataset Description The dataset represents synthetic but realistically calibrated electricity market data for selected West African countries. Key Features Macroeconomic indicators (GDP, inflation) Climate variables (temperature, humidity, reservoir levels) Power system capacity metrics Fuel price dynamics Market and reliability indicators 🧾 Data Dictionary Feature Description country West African country identifier demand_mw Electricity demand (MW) price_usd_kwh Electricity price (USD/kWh) outage_risk_score Probability-based index representing outage risk gdp_usd_billion Gross Domestic Product (Billion USD) inflation_pct Annual inflation rate (%) electrification_pct Percentage of population with electricity access temp_celsius_mean Mean ambient temperature (°C) humidity_pct Average humidity (%) reservoir_level_pct Hydropower reservoir level (%) installed_capacity_mw Total installed …