In recent years, in Mexico and in the world, surveys on victimization and perception of violence (ENVIPE, 2016) have become very important. This information is collected through a national sample collected by the Instituto Nacional de Geografia Estadistica e Informatica (INEGI, 2017), with the purpose of providing governments, and the authorities responsible for functions related to security and justice, information relevant on different aspects of this phenomenon, with the objective that decision making is made based on objective information. Unfortunately, most of these studies remain at the descriptive level and leave aside the multiple relationships that can occur between variables or sets of variables, whose understanding would give greater clarity about the phenomenon, by identifying factors, similarities or dissimilarities, dependency relationships or association, adjust models and make forecasts for intervention purposes. In order to identify some of these multiple relationships, a set of supervised methods of data mining, such as classification trees, regression trees and rule-based methods, are applied to the ENVIPE 2016 database (INEGI, 2016); in addition to some unsupervised methods such as Clustering. The procedures are executed in the software called WEKA, the models are adjusted using a training set, statistically significant attributes are identified and the models are adjusted. The selection of the models is done based on the criteria of goodness defined for them and later predictions are made in the cases in which it applies. The variables associated with victimization and perceptions of violence are identified, according to the association criterion, and predictions are made, in the probabilistic sense, about these phenomena. KEYWORDS: Victimization, Data Mining, Decision Tree, Clustering. MSC: 62-07 (62P25)