Stock Price Prediction Based on Joint LSTM and Fully Connected Layer

The significance of stock market prediction is extensive and crucial. It provides investors with insights into market trends, aiding them in making informed investment decisions, and assists businesses in understanding market expectations, contributing to government policies, financial analysis, and asset allocation decisions. In this paper, a deep learning approach is employed, utilizing long short term memory (LSTM) layers and fully connected layers to predict the stock price of a technology company. While achieving high accuracy, the study analyzes the effects of epochs and the number of LSTM layer neurons on prediction performance and accuracy. Specifically, time series data is grouped, processed through LSTM layers, optimized and dimensionally reduced with the help of fully connected layers for prediction and comparison with test data. Additionally, the paper examines the influence of the number of epochs and neuron count on the model by making modifications. This research opens new avenues for predicting future stock market trends and mitigating investment risks, offering an approach to prediction based on LSTM. Moreover, it provides recommendations for optimizing future models through an analysis of various parameters.