Effectiveness of Random Forest Model in Predicting Stock Prices of Solar Energy Companies in India

Abstract

The solar energy industry's positive impact on India's GDP is perceptible through increased investments, innovation, and enhanced energy security. As the nation continues to prioritize clean energy solutions, the solar sector stands as a key player driving both economic prosperity and environmental sustainability, aligning India with world wide determinations to battle climate change and encourage a greener future. As the Indian government continues to champion initiatives promoting renewable energy, Solar Energy Companies have seen unprecedented growth and have become increasingly attractive to investors seeking long-term, sustainable returns. This influx of interest, however, brings with it the challenge of navigating the volatile and dynamic nature of the stock market. In this context, forecasting the stock prices of solar energy corporations in India becomes a pivotal aspect of investment strategy for both institutional and retail investors. This paper targets to add to the prevailing body of knowledge by evaluating the efficacy of the Random Forest model, a machine learning technique known for its versatility and robustness, in forecasting the stock prices of top four Solar Energy Companies in India on the basis of market capitalization, by using the daily opening, high, low and closing stock prices ranging from 1st October, 2019 to 30th September, 2023 i.e. 4 years. The findings reveal that high Coefficient of Determination (R2) values for all companies, ranging from 0.9928 to 0.9939 is a clear indication of the model's ability to predict a substantial portion of the variance in each company's stock prices. But in case of Adani Green Energy Ltd a notably higher MSE and RMSE are exhibited, implying a greater degree of fluctuation in prediction accuracy compared to the other companies. On the other hand, all the selected solar energy companies display lower MAE values, indicating tightly clustered predictions around actual values.