An Improved Quantile-Point-Based Evolutionary

The international arab journal of information technology

article

Lei Liu¹  Zheng Pei¹  Peng Chen¹  Zhisheng Gao¹  Zhihao Gan¹  Kang Feng¹ 

[1] School of Computer and Software Engineering, Xihua University

关键词: Time series;    feature representation;    quantile segmentation points;    linear segmentation;    genetic algorithm;   

DOI  :  10.34028/iajit/19/6/4

学科分类：计算机科学（综合）

来源: Zarqa University

【 摘 要 】

Effective and concise feature representation is crucial for time series mining. However, traditional time series featurerepresentation approaches are inadequate for Financial Time Series (FTS) due to FTS' complex, highly noisy, dynamic and non- linear characteristics. Thus, we proposed an improved linear segmentation method named MS-BU-GA in this work. The criticaldata points that can represent financial time series are added to the feature representation result. Specifically, firstly, we proposea division criterion based on the quantile segmentation points. On the basis of this criterion, we perform segmentation of thetime series under the constraint of the maximum segment fitting error. Then, a bottom-up mechanism is adopted to merge theabove segmentation results under the maximum segment fitting error. Next, we apply Genetic Algorithm (GA) to the mergedresults for further optimization, which reduced the overall segment representation fitting error and the integrated factor ofsegment representation error and number of segments. The experimental result shows that the MS-BU-GA has outperformedexisting methods in segment number and representation error. The overall average representation error is decreased by 21.73%and the integrated factor of the number of segments and the segment representation error is reduced by 23.14%.

【 授权许可】

Unknown