台北市住宅竊盜犯罪地圖製作與犯罪區位分析

Abstract

住宅竊盜與犯罪發生地點息息相關，竊賊必須至某住宅行竊，犯罪行為具有空間相依性。本研究以台北市為研究地區，製作住宅竊盜犯罪地圖，辨識犯罪熱點，探討住宅竊盜的區位特性。由核密度推估圖與Getis-Ord G值犯罪地圖顯示，住宅竊盜犯罪熱點有從市中心向外擴展的趨勢。犯罪區位分析發現高教育程度人口比率、20∼60歲人口比率、相對地價殘差、建地密度、人口密度等預測因子與住宅竊盜率有統計上的顯著相關，其R2為0.316。由於高教育程度人口和相對地價殘差高可提升目標吸引性，建地密度和人口密度高會增加犯罪機會，台北市的住宅竊盜區位較著重於目標吸引性高和犯罪機會多。地理加權迴歸能反應空間變異情形，R2值自0.316提升至0.568，殘差總和從28.1下降至17.8。

Burglary is the most frequent crime in Taiwan nowadays and is closely related to geographical locations. This study is aimed to identify the hotspots of residential burglary in Taipei in 2000 and 2004 and to determine the locational characteristics of these hotspots. The locations of residential burglaries in Taipei were converted into a point map by matching the addresses reported to the police. The residential burglaries were not randomly distributed in Taipei, but concentrated in certain areas. Both burglary hotspots of Taipei City obtained by using kernel density and Getis-Ord G are concentrated around the city center and spread to the city outskirt. To understand the locational characteristics of high-intensity residential burglary areas, this study presented two kinds of regression models. Stepwise multivariate regression analysis showed that population with college or higher degrees, relative housing price, population of age 20-60, density of built-up area, and population density were significantly correlated with intensity of residential burglaries (R2 = 0.316). The results support the routine activity theory, which suggests burglaries occur in areas of higher daily activities, areas with more opportunities for burglaries and have less chance of being arrested. The other model is the geographically weighted regression model that had a R2 value of 0.568, higher than the global regression model.