ladymore/nigeria_electricity_demand

The model that is able to predict the demand of electricity per Distribution companies or zone in Nigeria # Nigeria Electricity Demand Predictor Predicts zonal electricity demand for Nigeria's power distribution companies, and gives national-level context on demand, generation, grid loss, and household access to electricity. ## Problem Context Nigeria's power grid faces chronic gaps between generation and demand, high transmission losses, and uneven household access to electricity across regions. Grid operators and Discos need reliable short-term demand forecasts to plan generation dispatch, reduce load-shedding, and manage grid stability. This project builds a regional demand forecasting pipeline covering three major distribution company zones: EKEDC/IKEDC (Lagos), IBEDC (Ibadan), and AEDC (Abuja). ## What's in this repo | File | Purpose | |---|---| | `nigeria_electricity_demand.py` | End-to-end pipeline: data preparation, data exploration, feature engineering, model training, and evaluation | | `app.py` | Streamlit app that loads the trained models and does live predictions | | `requirements.txt` | Python dependencies | | `models/` | Trained model bundles (`.joblib`) and generated data exploration outputs | | `README.md` | This file | ## Data Sources - **Global Electricity Demand and Generation Dataset** — national-level demand/generation (TWh) by country and year - **World Bank indicators** — electricity access rate (% of population) and transmission/distribution losses (%) - **Nigeria Electricity Data** (Excel) — household-level electricity access by state - **Zonal power consumption dataset** (`powerconsumption.csv`) — 10-minute interval consumption readings with weather covariates, used for the regional short-term forecasting models ## Pipeline Overview 1. **Data preparation** — merged national demand/generation figures with grid loss and access-rate data by year; disaggregates the most recent national net-delivered energy figure down to each Disco zone, weighted by that zone's share of households with electricity access. 2. **Feature engineering** — buil …