A Novel Technique to Minimising Mean Square Error in Stock Price Index Prediction Utilising Logistics Regression and LSTM Model

Abstract

The methodology that is recommended makes an attempt to anticipate and forecast changes in the price indices of the stock market for three specific equities that are traded on the stock market. The stock market is comprised of all of the many types of equities that are now being discussed. This paradigm makes an attempt to classify two unique types of classification algorithms, namely Long-Term Memory (LSTM) and Logistics Regression (LR). Long-Term Memory is an acronym for "Long-Term Memory," while Logistics Regression is an acronym for "LR." (LSTM). One of the criteria that is used to assess the performance of the models is the ability of both models to accurately anticipate the movement of an index that is traded on the Bombay Stock Exchange. (BSE). For the purposes of performing a study of the suggested structure for the projection of three stocks, an estimated total of thirty different participants were used. (AAPL, MSFT, and AMZN). When comparing the two models' levels of performance, it was found that the LR model (99.8%) performed substantially better than the LTSM model (72.3%) on average. This was noticed while conducting the comparison. (p0.05). When it comes to predicting stock indices by making use of the various parameters, the LR model performed noticeably better than the LTSM model.