["Transactions in GIS, Volume 30, Issue 4, June 2026. ", "\nABSTRACT\nInvestigating the complex relationship between urban environment and street crimes is crucial for understanding the mechanisms of crime occurrence. However, most studies treat the urban environment as homogeneous, overlooking the spatial heterogeneity of environmental influences on crime. To address this gap, this study introduces a novel framework to explore the heterogeneous influence of urban environmental characteristics on street crimes by integrating multi‐source urban data and employing the explainable machine learning method. First, environmental characteristics are quantified by integrating urban big data, such as points of interest and street view images. Second, explainable machine learning is employed to reveal the environment–crime relationship, which is reflected by Shapley additive explanation (SHAP) values. Finally, the urban environment is partitioned into homogeneous clusters using spatial clustering of SHAP values, enabling assessment of spatial variations of environmental influence on crime across different clusters. Experimental results reveal substantial spatial heterogeneity in the effects of environmental characteristics on street crimes across different locations. The proposed methodology can provide critical guidance for developing precise crime prevention strategies.\n"]