Predictive model based on Naive Bayes through Supervised Machine Learning and student dropout, in public Technological Education centers in the La Libertad region

This research expresses a predictive model to estimate students at risk of dropping out of school in public technological higher education centers in the La Libertad region. The model is based on the Naive Bayes classification algorithm in supervised learning machines guided by the CRISP DM methodology. The research is applied, descriptive, non-experimental and cross-sectional in design. The data is obtained from socioeconomic records, enrollments and historical notes to obtain the initial data set, after processing, the final data set is obtained. In the implementation, Python was used through Jupiter Notebook from Google Colaboratory. A part of the final data set was used to train, validate and another to evaluate the reliability of the model. An object of the algorithm is trained with the final set, and the predictive model is obtained. Once the model is generated, a prediction is made with the test data set and the reliability of the results is evaluated. With the expected results of the test data set, a degree of reliability of the obtained model of 93% is verified. To visualize the number of correct and incorrect patterns recognized by the model, the Confusion Matrix was used.