Istanbul Electricity Demand Forecast Using Neural Networks

This technical report presents a study on forecasting electricity demand in Istanbul using artificial neural networks. The objective is to optimize electricity demand forecasts to ensure efficient resource usage and minimize outages. The study evaluates the impact of seasonal variables such as temperature and humidity, as well as the influence of weekends and holidays on electricity consumption. A dataset from January 2021 to December 2023 was analyzed, utilizing NARX models and Prophet/LSTM hybrid models to enhance prediction accuracy. The findings indicate that factors like humidity, temperature, intra-day hours, and holiday periods significantly affect electricity consumption patterns. The report details the methodology for generating synthetic data using Generative Adversarial Networks (GANs) to address the lack of province-specific hourly data. The analysis includes exploratory data analysis (EDA) to understand variability in energy demand over time, revealing significant seasonal trends and consumption patterns that are crucial for energy producers and distribution companies.