期刊名称：Indian Journal of Computer Science and Engineering
印刷版ISSN：2231-3850
电子版ISSN：0976-5166
出版年度：2021
卷号：12
期号：2
页码：326-347
DOI：10.21817/indjcse/2021/v12i2/211202036
出版社：Engg Journals Publications
摘要：Financial sector is comprising of rising challenges and complexities. So, to handle time series prediction some recent artificial intelligent techniques are developed. In this paper, a technique named Simplex Method based Social Spider Optimization (SMSSO) is developed to predict time-series datasets like absenteeism at work, energy consumption, blog feedback data, and currency exchange rate. Performance of different artificial intelligence methods are checked with and without some feature selection techniques like ANOVA, Kruskal Walis, and Friedman test. Required features are selected that reduces the size of data to make easy analysis. Recent year data is associated in testing and the previous year’s data is used in training. The performance parameters during classification are Mean Square Error (MSE) value and time for execution which is divided into training and testing time. Results show 0.542 as minimum MSE in 1.142 sec of testing time when associated with ANOVA in currency exchange rate, 0.613 MSE in 2.102 sec of testing time when associated with Kruskal Walis in Blog feedback data, and 0.403 MSE in 1.367 sec of testing time when associated with Friedman test using electricity consumption and 0.210 of minimum MSE in 2.102 sec of testing time without any feature selection with blog feedback data in case of SMSSO-NN. The SMSSO-NN results are also compared with different classification algorithms like Rough Set Theory and Structured Singular Value (SSV).
关键词：Structured Singular Value (SSV); Root Mean Square Error (RMSE); Simplex Method based Social Spider Optimization (SMSSO); Kruskal Walis; Friedman test; Rough Set Theory; ANOVA.
