Examining Engineering Students’ Computational Thinking Skills with CART Analysis

Abstract

This study aims to examine the computational thinking skills (CTS) of engineering students through classification and regression trees (CART) analysis and to develop a predictive model using machine learning (ML) algorithms. The sample consists of 435 engineering students from the engineering faculties of three state universities located in Southeast and East Anatolia. A relational survey model is utilized, and data are collected via a personal information form and the CTS scale. Data analyses include two-stage cluster analysis, CART analysis and ML algorithms. Findings reveal that 71.43% of students possess high CTS, 12.70% are at a medium level and 15.87% are at a low level. The majority of engineering students exhibit moderate to high CTS. The variable department is identified as the most significant predictor of students’ CTS. Dimensions of computational thinking (CT) are analyzed using CART, identifying the most predictive variables for creativity, algorithmic thinking, cooperativity, critical thinking and problem-solving. ML algorithms, including AdaBoost, decision tree, multi-layer perceptron, naive Bayes and random forest, are employed to predict the CTS. The study’s findings and data patterns underscore the importance of targeted planning in engineering education to foster the effective development of CTS.