Predicting Stock Market Price: A Logical Strategy using Deep Learning

In time series data analysis, stock market prediction is particularly hard. In addition, for the best estimation of stock prices, proper tuning of the model is crucial. This research work uses the frequently used algorithms Long Short Term Memory, Extreme Gradient Boosting (XGBoost), Linear Regression, Moving Average, and Last Value model on more than twelve months of historical stock data to build up a prediction model for forecasting stock price. For the purpose of comparing among the models, the measurement of Mean Absolute Percentage Error (MAPE) is used and it is observed that the LSTM method exceeds all the other methods with a MAPE of 0.635. Furthermore, the highest error rate among the five models is found for Moving Average for our case.