A novel artificial immune system-based approach for mining associative classification rules with stock trading data. (2017)

• Record Type: Journal Article
• Authors: Ghodsi, Mahsa Mahboob; Zandieh, M.
• Is Part Of: International journal of innovative computing and applications. Volume 8:Number 3(2017)
• Page Start: 149
• Page End: 161
• Publication Date: 2017
• Subjects: stock market prediction -- data mining techniques -- associative classification rule mining -- ACR -- artificial immune system -- AIS; Evolutionary computation -- Periodicals; Neural networks (Computer science) -- Periodicals; Genetic programming (Computer science) -- Periodicals; Biologically-inspired computing -- Periodicals; Swarm intelligence -- Periodicals; Quantum computers -- Periodicals; 006.3
• Languages: English
• ISSNs: 1751-648X

Abstract

Stock market prediction with high accuracy has always been an interesting subject for most investors and professional analysts. Data mining techniques are providing great aid to extract interesting and hidden knowledge from datasets. Financial data mining tools assist investors in their investment decisions, thereby reducing their investment risks. Associative classification rule mining is a promising approach in data mining that utilises the association rule discovery techniques to construct classification systems, also known as associative classifiers. This paper aims to develop an intelligent transaction system based on associative classification rule mining (ACR) and phenotypic artificial immune system (AIS) which discovers trading rules from numerical indicators. A new fitness function as a different measure of quality for quantitative association is suggested considering interestingness of rules. Based on the empirical studies on the top eight companies in the S&P 500 stocks, observed results demonstrate the superior prediction accuracy over the genetic algorithm based technique and the 'buy and hold' strategy.