A case-based reasoning approach for pattern detection in Malaysia rainfall data. (2015)

• Record Type: Journal Article
• Authors: Alshareef, Almahdi Mohammed; Bakar, Azuraliza Abu; Hamdan, Abdul Razak; Abdullah, Sharifah Mastura Syed; Alweshah, Mohammed
• Is Part Of: International journal of big data intelligence. Volume 2:Number 4(2015)
• Page Start: 285
• Page End: 302
• Publication Date: 2015
• Subjects: data mining -- CBR -- case-based reasoning -- big data -- pattern detection -- rainfall data -- data streams -- data segmentation -- classification -- pattern discovery -- Malaysia -- precipitation data -- rainfall patterns -- adaptive sliding window algorithm -- ASWA; Big data -- Periodicals; 005.705
• Languages: English
• ISSNs: 2053-1389

Abstract

A pattern detection algorithm for a rainfall time series dataset is proposed. However, the adaptive sliding window algorithm (ASWA) and the improved case-based reasoning (CBR) approach namely: the data segmentation phase of pattern detection adapts the classical sliding window algorithm (SWA) to compute change points in time series rainfall data, where the ASWA improves the SWA with respect to two parameters, namely, error and window size. In pattern detection phase, the goal is to detect the class labels of the rainfall dataset. Two phases of CBR are improved: the retrieval phase is adapted to compute rainfall sequences with unequal window sizes and the revise phase is improving classification accuracy. Experiments show that the proposed ASWA generates a smaller number of windows compared to the classical SWA while the proposed improved CBR approach detects the patterns of segmented data from which experts can determine the class labels for the patterns.