5. Analysis using clustering techniques of the eating habits of students at the Technological University of the Mixteca, Oaxaca, Mexico.

Abstract

The K-Prototype unsupervised clustering algorithm was applied to a mixed dataset examining students' eating habits at the Technological University of the Mixteca (UTM) to assess the accuracy of students' self-perceptions regarding their diets. A survey was conducted with 121 students, incorporating both quantitative and qualitative variables. Variables analyzed included number of meals per day, gender, junk food consumption, alcohol or drug use, health perception, age, and perceived monthly income. The CRISP-DM methodology, a widely recognized framework for data analysis, guided the process through the phases of process. The phases of the CRISP-DM methodology are: understanding, data understanding, data preparation, modeling, evaluation, and deployment. The K-Prototype algorithm was selected for the modeling phase due to its capacity to handle both categorical and numerical data simultaneously. Findings indicate that women with healthy habits are consider themselves as unhealthy, whereas men with unhealthy habits are classified as healthy.