Profiles of Chilean students according to academic performance in mathematics: An exploratory study using classification trees and random forests

• Profile characterization for Chilean students based on performance in mathematics.
• Analysis made with classification and regression tree (CART) and random forest.
• Parents’ high educational expectations is key variable for best performance.
• Type of school and the index of mathematical abilities influence good performance.

Attempting to deepen the understanding of factors that explain student performance, this study seeks to identify and characterize profiles of Chilean students based on academic performance in mathematics. As analytical method, statistical techniques known as random forest (RF) and classification and regression tree (CART) were used to identify groups of eighth-grade elementary students according to their performance in 2011 test, using features related to individual and family behavior. The analysis was performed with a database provided by the Education Quality Measurement System of Chile. Results show that “parents’ educational expectations” (42.7%) is key factor to obtain the best children's performances. Additionally, the analysis showed that the “type of school” (26.6%) and the “index of mathematical abilities” (26.1%) influence good performance.