A hybrid SOFM-SVR with a filter-based feature selection for stock market forecasting Huang, C. L. & Tsai, C. Y. Expert Systems with Applications 2008

Introduction   Stock market price index prediction is regarded as a challenging task of the finance.   Support vector regression (SVR) has successfully solved prediction problems in many domains, including the stock market.

Introduction   filter-based feature selection to choose important input attributes   SOFM algorithm to cluster the training samples   SVR to predict the stock market price index   Using a real future dataset – Taiwan index futures (FITX) to predict the next day’s price index

Introduction   SOFM+SVR : to improve the prediction accuracy of the traditional SVR method and to reduce its long training time,   SOFM+SVR+filter-based feature selection : improvement in training time, prediction accuracy, and the ability to select a better feature subset is achieved.

SVR   Unlike pattern recognition problems where the desired outputs are discrete values (e.g., Boolean)   support vector regression (SVR) deals with ‘real valued’ functions

Self-organizing Feature Maps; SOFM

SOFM 12 34

Training the SOFM-SVR model  1.  1. Scaling the training set   2.Clustering the training dataset   3.Training the Individual SVR Models for Each Cluster

Training the SOFM-SVR model

Parameters Optimization   setting of the SVR parameters can improve the SVR prediction accuracy   Using RBF kernel and ε-insensitive loss function, three parameters, C, r, and ε, should be determined in the SVR model   The grid search approach is a common method to search for the C, r, and ε values.

Grid Search Approach

Evaluating the SOFM-SVR model with test set   Scale the test set based on the scaling equation according to the attribute rage of the training set   Find the cluster to which the test sample in the test set   Calculate the predicted value for each sample in the test set   Calculate the prediction accuracy for the test set

SOFM-SVR model

SOFM-SVR combined with filter- based feature selection   X is Certain input variable (i.e. feature)   Y is response variable (i.e. label)   n is the number of training samples

SOFM-SVR filter-based feature selection

Performance measures   A i is the actual value of sample i   F i is a predicted value of sample i   n is the number of samples.

Experimental data set

SOFM-SVR with various numbers of clusters in dataset #1

Accuracy measures with various numbers of clusters

Wilcoxon sign rank test Wilcoxon sign rank test on the prediction errors for the SOFM-SVR with various numbers of clusters

Results of SOFM-SVR using three clusters

Results of SOFM-SVR with selected features

Original Feature VS. Original Feature  Original Feature Wilcoxon sign rank test

Important Feature   MA10: 10-day moving average.   MACD9: 9-day moving average convergence/ divergence.   +DI10: directional indicator up.   -DI10: directional indicator down.   K10: 10-day stochastic index K   PSY10: 10-day psychological line.   D9: 9-day stochastic index D

Relative importance of the selected features

Wilcoxon sign rank test: SOFM-SVR vs. single SVR

MAPE comparison: SOFM-SVR vs. single SVRs.

Training time comparisons: SOFM- SVR vs. single SVRs.

Conclusion   Hybrid SOFM-SVR with filter based feature selection to improve the prediction accuracy and to reduce the training time for the financial daily stock index prediction   Further research directions are using optimization algorithms (e.g., genetic algorithms) to optimize the SVR parameters and performing feature selection using a wrapper-based approach that combines SVR with other optimization tools

Thank You