Prediction of Complex Stock Market Data Using an Improved Hybrid EMD-LSTM Model

Because of the complexity, nonlinearity, and volatility, stock market forecasting is either highly difficult or yields very unsatisfactory outcomes when utilizing traditional time series machine learning techniques. To cope with this problem improve complex market’s prediction accuracy, we propose a new hybrid novel method that based on version EMD deep technique known as long-short memory (LSTM) network. The precision proposed ensemble evaluated using KSE-100 index Pakistan Stock Exchange. Using uses Akima spline interpolation instead cubic interpolation, noisy data are first divided into multiple components technically intrinsic mode functions (IMFs) varying from high to low frequency single monotone residue. correlated sub-components then used build LSTM By comparing model other models such support vector (SVM), Random Forest, Decision Tree, its performance thoroughly evaluated. Three alternative statistical metrics, namely root means square error (RMSE), mean absolute (MAE) percentage (MAPE), compare aforementioned empirical results show suggested Akima-EMD-LSTM beats all taken consideration for study therefore recommended an effective non-stationary nonlinear financial data.