Web Application Based on Data Mining using the Naive Bayes Technique for the Prediction of childhood obesity in Public Hospitals of Lima

Abstract

Childhood obesity is a disease that currently causes much concern worldwide and causes different comorbidities in children such as diabetes and respiratory disorders, in addition to being a risk factor for COVID19. For these reasons, there are several investigations that seek to predict this disease using different Data Mining techniques such as Decision Trees, Logistic Regression, Neuro-Fuzzy Systems, among others. The present research work makes the prediction that a child under 5 years old will suffer from obesity in the future using the Naive Bayes technique; The data set to implement the model had 770 records and 27 variables extracted from the e-Qhali application. The tests were performed on 317 records obtaining a model with 72% precision and 93% sensitivity, and when comparing the Naive Bayes technique with other classification techniques such as Logistic Regression, Random Forest and SVM, it reached the highest percentage of sensitivity.