How iakashmall/load-forecasting- works
This is a power-load forecasting toolkit. An analyst loads the hourly CSV, normalizes and resamples it, engineers calendar, weather, lag, rolling, and Fourier features, then runs hourly ML/deep-learning or daily statistical/STL forecasts. Forecasts can be persisted, visualized through EDA plots, and evaluated with forecasting metrics. The dashboard directory is empty in the supplied tree, so no dashboard behavior is inferred. Several model internals and orchestration paths were only partially sampled; optional persistence and tuning links are therefore shown conservatively.
Components 21
Data Preparation
- Power Load CSV
data/PowerLoad_Dataset.csv - Data Loader
data/data_loader.py - Feature Pipeline
data/preprocessing.py - Daily Aggregation
data/data_loader.py
Exploratory Analysis
- EDA Runner
eda/exploratory_analysis.py - EDA Plots
outputs/plots
Hourly Forecasting
- XGBoost Forecaster
ml_models/xgb_model.py - Random Forest Forecaster
ml_models/random_forest.py - LSTM Forecaster
dl_models/lstm.py - XGBoost Prediction
ml_models/predict_xgb.py
Daily Forecasting
- ARIMA Forecaster
statistical_models/Arima.py - SARIMAX Forecaster
statistical_models/Sarimax.py - STL Forecaster
ml_models/stl_decomposition.py
Evaluation Outputs
- Forecast Metrics
evaluation/metrics.py - XGBoost Model
ml_models/xgboost_model.pkl - ARIMA Model
statistical_models/arima_model.pkl - SARIMA Model
statistical_models/sarima_model.pkl - Prediction Results
outputs/predictions/xgb_predictions.csv - Metric Results
outputs/metrics/xgb_metrics.json - Feature Importance
outputs/feature_importance/xgb_feature_importance.csv
Other components
- Analyst
Connections 24
- Analyst → Data Loader: loads data
- Power Load CSV → Data Loader: reads CSV
- Data Loader → Feature Pipeline: provides hourly data
- Data Loader → Daily Aggregation: aggregates daily
- Analyst → EDA Runner: runs analysis
- EDA Runner → Data Loader: loads data
- EDA Runner → EDA Plots: saves plots
- Feature Pipeline → XGBoost Forecaster: trains features
- Feature Pipeline → Random Forest Forecaster: trains features
- Feature Pipeline → LSTM Forecaster: builds sequences
- Analyst → XGBoost Prediction: runs prediction
- XGBoost Prediction → Data Loader: loads data
- XGBoost Prediction → Feature Pipeline: builds features
- XGBoost Prediction → XGBoost Forecaster: requests forecast
- XGBoost Forecaster → XGBoost Model: loads model
- XGBoost Forecaster → Prediction Results: writes predictions
- Daily Aggregation → ARIMA Forecaster: provides daily series
- Daily Aggregation → SARIMAX Forecaster: provides daily data
- Daily Aggregation → STL Forecaster: provides daily series
- ARIMA Forecaster → ARIMA Model: persists model
- SARIMAX Forecaster → SARIMA Model: persists model
- ARIMA Forecaster → Forecast Metrics: scores forecasts
- Forecast Metrics → Metric Results: writes metrics
- XGBoost Forecaster → Feature Importance: writes importance