Mexico has steadily increased homicide rates over the past few decades. To better understand this phenomenon, we examine homicides, victims, socioeconomic context, and weather conditions by focusing on two critical dimensions of crimes: space and time. We implement diverse regression models considering spatial (Geographically Weighted Regression) and spatio-temporal (Geographical and Temporal Weighted Regression) perspectives. These models are compared to general linear regression models (Ordinary Least Squares Regression, Generalized Least Squares, and Weighted Least Squares). Our findings highlight the importance of specific socioeconomic factors (e.g., educational backwardness, food insecurity, or poor-quality housing) and weather conditions in accounting for differences in homicides and victims. Furthermore, our experiments demonstrate that spatio-temporal models provide a better fit compared to general regression models. Our results provide valuable insights into the issue of homicides in Mexico. This work can help perform evidence-based interventions and policy-making at national and state levels. The outcomes emphasize the necessity of addressing socioeconomic disparities and considering weather conditions when tackling homicide issues in Mexico.