Outlier Detection Using a New Hybrid Approach Based on Group Weighted K-Mean and Greedy Method on Mixed Dataset

Provided by: Creative Commons
Topic: Big Data
Format: PDF
Outlier detection is currently very active area of research in data mining. An outlier is a pattern which is dissimilar with respect to the rest of the patterns in dataset. Proposed method for outlier detection uses hybrid approach. Purpose of approach is first to apply the clustering algorithm that is Group Weighted K-Means (GWK-Mean) which partition the dataset into number of groups and second using greedy algorithm for detect outliers. The principal of outliers finding depend on the threshold. Threshold is set by user. Entropy is also used to measure of disorder present in a system. The hybrid approach, two techniques are combined to improve efficiently find the outlier from the dataset.

Find By Topic