Real Time Energy Stock Forecasting Using a Hybrid AI and CI Framework with IoT Driven Big Data Analytics

Abstract

Energy stock price forecasting is difficult due to the fact that these market prices are influenced not only by the financial trends but also by the real world energy consumption, production, and supply fluctuations. Traditional models, which is based primarily on historical stock data, do not frequently succeed in representing these complexities. To face this, here we present the Computational Intelligence IoT Big Data (CI IoT BD) framework, which combines real time IoT signals, Big Data infrastructures, and hybrid computational intelligence methods. The framework includes the collection of data by smart meters, industrial sensors, and renewable plants, data processing using Hadoop and Spark, and the hybrid intelligence solution, which is the combination of LSTM networks, fuzzy logic, and genetic algorithms. Five years of IoT and financial data (2018 2023) were used to train the model, which was used to compare against ARIMA, GARCH, Random Forest, and standalone LSTM. Results: The results show that the data is improved significantly, RMSE has been improved to 9.44, MAPE is 4.87%, and Directional Accuracy is 81.3%, which is better than all baselines. The results show that IoT signals act as early warning indicators of stocks' movements and improve the accuracy and robustness during the periods of volatility. Aside from its technical contributions, this framework can assist investors in obtaining improved trading information, energy companies in coordinating operations with the market, and policymakers in developing sustainable energy finance.