OPTIMASI HEATMAP MENGGUNAKAN K-PROTOTYPE CLUSTERING DALAM PENGELOMPOKAN DATA KASUS PEMBUNUHAN BERANTAI

ANANDA, MUHAMMAD AL-AZIZ (2019) OPTIMASI HEATMAP MENGGUNAKAN K-PROTOTYPE CLUSTERING DALAM PENGELOMPOKAN DATA KASUS PEMBUNUHAN BERANTAI. Undergraduate Thesis thesis, Institut Telkom Purwokerto.

	Text COVER.pdf - Accepted Version Download (476kB)
Preview	Text ABSTRACT.pdf - Accepted Version Download (4kB) | Preview
Preview	Text ABSTRAK.pdf - Accepted Version Download (5kB) | Preview
Preview	Text BAB I.pdf - Accepted Version Download (169kB) | Preview
Preview	Text BAB II.pdf - Accepted Version Download (1MB) | Preview
Preview	Text BAB III.pdf - Accepted Version Download (418kB) | Preview
	Text BAB IV.pdf - Accepted Version Restricted to Registered users only Download (836kB)
Preview	Text BAB V.pdf - Accepted Version Download (6kB) | Preview
Preview	Text DAFTAR PUSTAKA.pdf - Accepted Version Download (126kB) | Preview

Abstract

ABSTRACT The presence of serial killers is a worrying problem in the community. On September 4, 2016, it was recorded in the Serial Killer Definition by the FBI that the serial killer had the highest IQ of 186 so more efforts were needed to catch him. The dragnet system that was initiated by David Canter was proven to increase the chances of the police to arrest perpetrators with a projected heatmap into the map of the area. Heatmap itself is visual that is formed from data that is projected in a field and will have a hot area following the density of points in that area. However, this method itself has problems in the validity of input and point setting in serial murder cases. Based on murder data obtained from the Murder Accountability Project, it has many problems such as damaged data items. K-Prototype Clustering method is applied to find new heatmap patterns with projections into 2D fields using Google Colab (python programming language) and RStudio (R programming language). The heatmap optimization stage using the k-prototype method is to preprocessing data, and comparison of 2D heatmap projections before and after clustering. The results of the research conducted show the similarity of grouping data through heatmap visualization and accuracy generated from the cluster formed. The Silhouette Index score generated by K-Prototype from heuristic feature selection is 0.1792507 at K = 6, and feature selection via PCA and MCA methods with a score of 0.1882191 at K = 10. The accuracy of the two clusters is 51%, visualization of the heatmap results in significant grouping changes. It can be concluded that clustering is the right method applied in classifying serial killer cases on murder data. Keyword: Data Mining, Geographic Profiling, Heatmap, K-Prototype Clustering, Serial Killer.

Actions (login required)

View Item