An Improved Log Loss Stock Market Prediction Using Novel Long Short Term Memory Algorithm in Comparison with Support Vector Machine Algorithm

Abstract

This work provides a comparative study of improved log loss stock market values using a novel long short term memory algorithm (LSTM) and support vector machine algorithm (SVM). Novel Long Short Term Memory (N = 10) and support vector machine (SVM) (N = 10) where iterated to improve log loss stock market predicted values in stock price prediction.Simulation was done by varying NLSTM and support vector machine parameters to optimize the pH. Sample size is calculated using Gpower for two groups and 20 samples were used in total for this work.LSTM has notably better accuracy percentage (68.11) compared to SVM accuracy (52.46). Statistical significance difference between long and short term memory and support vector machine algorithm was found to be 0.023 (p<0.05).Long short term memory algorithm provides better results in log loss stock market values than support vector machines.