Predictive Modeling for Early Detection of High School Dropouts Using Machine Learning Techniques

Abstract

This research paper presents an innovative approach to developing a predictive model for early identification of high school dropouts using machine learning algorithms. The study analyzes the National Center for Education Statistics dataset to create an effective dropout detection system. To address the challenge of high dimensionality in the dataset, principal component analysis is applied to reduce its complexity. The study compares the performance of different machine learning methods, including a multi-layer artificial neural network, k-nearest neighbors, a support vector machine with a radial basis function kernel, and a support vector machine with a polynomial kernel. The objective is to determine the most accurate classifier for predicting dropout risk. The experimental results highlight the neural network as the top-performing classifier, with statistically significant differences compared to k-nearest neighbors. These findings contribute to developing proactive measures and interventions to prevent high school dropouts and enhance educational outcomes.