地理的プロファイリングにおける拠点推定モデルの精度比較—犯行地点数の影響の検討—  [in Japanese] Comparing the accuracy of geographic profiling models and effects of the number of crime locations  [in Japanese]

Access this Article

Search this Article

Author(s)

Abstract

<p>本研究では,拠点推定モデルの精度について,犯行地点数を考慮した検討を行った。分析では,15地点以上の住宅侵入盗を行った55名分のデータを使用し,サークル仮説中心,空間平均,CMD(center of minimum distance: 最小距離中心)の各モデルについて,3地点から15地点までの各犯行地点数における誤差距離(犯人の推定居住地と実際の居住地の間の直線距離)を比較した。多重比較を行った結果,犯行地点数が7地点以上の場合に,CMDにおける誤差距離がサークル仮説中心よりも短かった。結果から,犯行地点数がモデルの推定精度に与える影響はモデルによって異なることが示唆された。</p>

<p>Effects of the number of crime locations on geographic profiling models were investigated. The data of 55 serial residential burglaries comprised of up to 15 burglaries committed by each burglar were sampled. The accuracy of 3 geographic profiling models: center of the circle, spatial mean, and CMD models were compared in each of 3 to 15 crime locations. The error distance was used for measuring the accuracy of the models. Multiple comparisons showed that CMD method had shorter error distances than center of the circle method when seven, or more crime locations were used. These results suggested that the number of crime locations differentially affected the accuracy of each model.</p>

Journal

  • The Japanese Journal of Criminal Psychology

    The Japanese Journal of Criminal Psychology 54(1), 41-51, 2016

    The Japanese Association of Criminal Psychology

Codes

  • NII Article ID (NAID)
    130005465242
  • NII NACSIS-CAT ID (NCID)
    AN00206218
  • Text Lang
    JPN
  • ISSN
    0017-7547
  • NDL Article ID
    027666153
  • NDL Call No.
    Z2-85
  • Data Source
    NDL  J-STAGE 
Page Top