Data Driven Approach for Analysing the Relationship Between Weather Patterns and Energy Consumption in Residential Buildings for Improved Energy Management

Residential buildings are the most significant providers of global energy consumption. Hence, studying the weather patterns and energy use in residential buildings remains vital for effective energy management strategies. This work explores a data-driven approach to analyse the complex interplay between weather and energy consumption in residential settings. Traditional methods for understanding building energy use often rely on physical models or occupant surveys, which can be time-consuming and expensive. This research addresses the need for a more efficient and comprehensive approach to analyse the effect of weather patterns on energy consumption. Following deep learning sequence models were used to most accurately predict the energy consumption - LSTM (Long Short Term Memory), Bi-LSTM (Bidirectional LSTM) and Attention Based LSTM (ABLSTM) and find the ABLSTM model most accurate and suited for our requirement with RMAE value as 0.3086, R2 as 0.9990. This research provides individuals and communities with the knowledge to make informed decisions about their energy usage, potentially leading to reduced energy costs through optimized consumption and can contribute to improved energy security.