Sexual crime is a critical global social problem. There remains a critical knowledge gap concerning whether and to what extent sexual crimes in public outdoor spaces can be influenced by landscape morphology of green spaces. This missing knowledge hinders the effective use of green spaces to reduce sexual crimes in these public settings. To address this issue, we collected a dataset comprising 5,155 cases of sexual crimes that occurred in public outdoor spaces in the United States from August 2021 to July 2022. A random forest model was employed to examine the statistical relationships between landscape morphology and sexual crimes. Additionally, we utilized the Shapley Additive Explanations (SHAP) model to quantify the interaction effects of landscape morphology with socioeconomic and demographic characteristics. This study yields three key findings: (1) Both the proportion and configuration factors of landscape morphology may significantly influence the sexual crime probability. (2) The relationships between landscape morphology and sexual crimes are nonlinear, and threshold values for the satisfactory dose and the preferred dose of green spaces can be identified. (3) There are significant interaction effects between landscape morphology with socioeconomic and demographic characteristics, emphasizing the importance of prioritizing green space interventions in socioeconomically disadvantaged areas. Lastly, through summarizing the findings of this study and previous research, we propose the Landscape-Sexual Crime Model (LSCM), which advocates for further research to explore effective strategies for using green spaces to reduce sexual crimes.

中文翻译：

---

减少公共户外空间性犯罪的自然剂量：提出景观-性犯罪模型


性犯罪是一个严重的全球社会问题。关于公共户外空间中的性犯罪是否以及在多大程度上会受到绿色空间景观形态的影响，仍然存在严重的知识差距。这种知识的缺失阻碍了有效利用绿色空间来减少这些公共场所的性犯罪。为了解决这个问题，我们收集了2021年8月至2022年7月美国公共户外空间发生的5155起性犯罪案件的数据集。采用随机森林模型来检验景观形态与性犯罪之间的统计关系。此外，我们利用沙普利加法解释（SHAP）模型来量化景观形态与社会经济和人口特征的相互作用效应。本研究得出三个主要发现：（1）景观形态的比例和配置因素都可能显着影响性犯罪概率。 (2)景观形态与性犯罪之间的关系是非线性的，可以确定绿地满意剂量和优选剂量的阈值。 (3)景观形态与社会经济和人口特征之间存在显着的交互作用，强调了优先考虑社会经济弱势地区绿地干预的重要性。最后，通过总结本研究和以往的研究结果，我们提出了景观-性犯罪模型（LSCM），主张进一步研究探索利用绿色空间减少性犯罪的有效策略。