Applying DevOps Practices for Machine Learning: Case Study Predicting Academic Performance

This paper presents the application of DevOps practices in the development and implementation of a machine learning model to predict the academic performance of students enrolled in a Higher Education Institution in Ecuador. The study encompasses three key phases: 1) Dataset preparation to build and validate the model, 2) Experimental evaluation of machine learning algorithms for academic performance prediction, and 3) Implementation of DevOps practices to design and build a software application using Microservices architecture, that supports the prediction model and visualize the results to take decisions. The dataset employed for the model and software application incorporates socio-demographic data and academic records of undergraduate students within the Information Technology career. Logistic Regression, Random Forest, and Neural Network algorithms are applied directly to the student dataset. Compared with other algorithms, experimental results present favorable outcomes with Random Forest, particularly in precision, sensitivity, and f-score metrics.