The main objective of this project is to develop a comprehensive framework for analyzing, predicting, and optimizing energy consumption in a smart home environment using weather-driven data. By the use of advanced time-series analysis techniques and ML models, this project aims to recognize patterns in energy usage, detect anomalies, and provide usable insights for efficient usage of energy. The end goal is to create a system capable of predicting energy consumption and offering optimized device configurations that reduce costs and environmental impact without affecting household comfort.
This project highlight key findings into the dynamics of energy consumption and the efficacy of predictive and optimization models. We have used the VAR, Prophet and LightGBM Regressor model in predicting energy as these models acheived a lower mean absolute error (MAE) score and showed better predictive performance compared to other models like VARMAX, LSTM and Support Vector Machine.
Dependency between home appliances and weather conditions
Total energy consumption of various household appliances by day
Total Energy consumption (actual vs predicted by the VAR model)
Total Energy consumption (actual vs predicted by the Prophet model)
Total Energy consumption (actual vs predicted by the LightGBM Regressor model)
SHAP model showing the impact of features on the prediction model
This is the Link: Smart Home Energy Consumption and Weather Data