Time Series Forecasting for Sustainable Power Consumption Prediction

This project presents a time series forecasting model for power consumption prediction using Recurrent Neural Networks with Long Short-Term Memory (LSTM). It addresses the limitations of traditional statistical models like ARIMA in handling non-stationary, unevenly spaced power data. By processing historical consumption data and applying seasonal decomposition, the model learns temporal patterns to accurately forecast future demand. The system shows strong performance, explaining about 86% of data variance, and offers business value by aiding resource optimization, infrastructure planning, and energy efficiency in smart grids.