Neural Networks-Based Forecasting Platform for EV Battery Commodity Price Prediction

Abstract

This study explores the impact of green energy-based economies on the growing use of electric vehicle (EV) batteries in transportation and electronic devices. Despite the environmental benefits, concerns have emerged regarding the supply, pricing, and volatility of raw materials used in battery manufacturing, exacerbated by geopolitical events such as the Russian-Ukrainian war. Given the high uncertainty surrounding EV commodity materials, this research aims to develop forecasting tools for predicting the prices of essential lithium-based EV battery commodities, including Lithium, Cobalt, Nickel, Aluminum, and Copper. The study builds on previous research on commodity price forecasting. Using Neural Networks such as LSTM that run using analytics platforms like RapidMiner, a robust and accurate models is able to be produced while require little to no programming ability. This will solve the needs to produce advanced predictions models for making decisions. As the results from the research, the models that are produced are successful in generating good prediction models, in terms of RMSE of 0,03 – 0,09 and relative errors of 4-14%.