Predicting the Tehran Stock Exchange Index Using Support Vector Regression; Based on the Dimension Reduction Technique

Stock markets play a significant role in organizing modern economic systems. Several research projects have been performed in the field of prediction using intelligent techniques. Considering that the accuracy of these techniques is significantly affected by their input features, one of the improvements made in the use of intelligent models, in addition to the application of hybrid models, is the use of dimensionality reduction as a preprocessing for the prediction model. In this study, in order to predict the daily index of the Tehran Stock Exchange, two methods of dimensional reduction (selection and extraction) are used simultaneously to select appropriate features as model inputs. Hence, the MRMR-MID algorithm is used to select features and the PCA algorithm is used to extract features. Then, support vector regression is used as a prediction model. Finally, an algorithm for selecting suitable features is proposed as ISF_ MID, according to the results obtained from the analysis of the use of dimensional reduction techniques in the prediction model. The results show that with the proposed method, with 7 selected features, it is possible to achieve high accuracy in predicting the daily index of the Tehran Stock Exchange. It should be noted that the studied models were evaluated in the implementation stage by the k-fold cross-validation method. MAE, MSE, and RMSE criteria are also used to evaluate the performance of these models.