This research involves the study, design, and implementation of an early warning model based on machine learning techniques to detect university dropout in technical programs at the Universidad de las Fuerzas Armadas ESPE in Ecuador. To achieve the proposed objectives, a group of technical experts with several years of experience in university teaching was assembled. These professionals came from diverse fields such as early childhood education, computer science, and economics, which enabled the design of an instrument that considered both socioeconomic and academic dimensions to understand better the reality of students at risk of dropping out. After being validated by academic peers, this instrument was administered through surveys to students enrolled in technical programs. The data collected underwent a cleaning process, after which machine learning techniques were applied to analyze the dropout phenomenon. The findings confirmed that several issues must be addressed early, at the time students enter the university. For the development of the early warning model, the CRISP-DM methodology was employed, which proved highly useful in achieving the research objective.