Smart home energy optimization system

The proposed system outlined a scientific project aimed at developing a machine learning-based system to optimize home energy usage and optimization. The proposed system leverages occupancy patterns, weather forecasts, and energy consumption data to create predictive models to recommend energy-efficient actions to homeowners. By utilizing advanced machine learning techniques, this research study aims to contribute sustainable energy practices and reduce energy costs for homeowners while minimizing environmental impact. The proposed system was developed to analyze the data using exploratory data analysis approaches. Pre-processing approaches are applied to prepare the data for model development. Weather correlations are identified with the usage of energy for home appliances. Groups are created based on the division of the date column data such as month-wise, weekly, daily, and hourly. The algorithms for the data forecasting used moving average, persistence algorithm, ARIMA, auto ARIMA, LSTM univariate and LSTM multivariate. The performance of the proposed system was evaluated by showing the graphical representation, which was very satisfactory. The LSTM multivariate algorithm outperforms the smart home energy optimization dataset compared to other algorithms. The outcome in the graphical representation of the evaluation shows much satisfaction.