Stock Price Prediction by using Machine Learning Techniques: a Study of TCS Ltd

Prediction of stock prices is challenging, but now stock price prediction is becoming very popular among researchers. Exact prediction of the stock market is not feasible, but the present study is an attempt to understand stock price prediction with special attention to TCS Ltd. The key objective of the present study is to forecast the high price of TCS stock on a daily basis. To achieve the objective successfully, this study comprises the use of historical time-based data for 15 years, from 2009 to 2023. (12 years of data for training purposes and 3 years of data for testing purposes are significantly categorized). Day-wise high prices have been significantly predicted by using open prices, low prices, and close prices. This study comprises the use of Multi Linear Regression (MLR), Support Vector Machine (SVM) with Linear Kernel, Ploy Kernel, and RBF Kernels, Multilayer Perceptron (MLP 2, 2), and XGBOOST algorithms. To access the accuracy of results from different models, Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) also been significantly calculated. Multi Linear Regression (MLR) model showed more accurate results for the prediction of day-wise high prices as compared to other algorithms. Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) of MLR found 11.57 and 14.69 respectively, the lowest of all the algorithms. Traditional regression tools are not as effective as these algorithms because these algorithms can provide more accurate and precise results to serve the interests of investors. This study concluded with suggestions to the investors while deciding whether to hold the stocks or sell them on a particular day.