Performance Comparison of Attribute Set Reduction Algorithms in Stock Price Prediction - A Case Study on Indian Stock Data

• Published in: Swarm, Evolutionary, and Memetic Computing pp. 567 - 574
• Main Authors: Sivakumar, P. Bagavathi, Mohandas, V. P.
• Format: Book Chapter
• Language: English
• Published: Berlin, Heidelberg Springer Berlin Heidelberg
• Series: Lecture Notes in Computer Science
• Subjects: attribute set reduction algorithms; computational intelligence; Data mining; Evolutionary algorithms; Financial time series; Stock trend prediction
• ISBN: 3642175627; 9783642175626
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-642-17563-3_66

Stock price prediction and stock trend prediction are the two major research problems of financial time series analysis. In this work, performance comparison of various attribute set reduction algorithms were made for short term stock price prediction. Forward selection, backward elimination, optimized selection, optimized selection based on brute force, weight guided and optimized selection based on the evolutionary principle and strategy was used. Different selection schemes and cross over types were explored. To supplement learning and modeling, support vector machine was also used in combination. The algorithms were applied on a real time Indian stock data namely CNX Nifty. The experimental study was conducted using the open source data mining tool Rapidminer. The performance was compared in terms of root mean squared error, squared error and execution time. The obtained results indicates the superiority of evolutionary algorithms and the optimize selection algorithm based on evolutionary principles outperforms others.