Performance-based evaluation of computational thinking skills using machine learning;

Abstract

This thesis study is an evaluation tool that measures the user's Computational Thinking (CT) according to its performance on Block-Based Programming Languages (BBPL) by using Machine Learning Techniques. The evaluation tool makes an evaluation for Machine Learning Model and each of the Computational Thinking seven fields. Via this developed tool, the objective is to product an alternative to the limited evaluation tools on Block-Based Programming Languages. Literature research was brought out in two sides as Machine Learning and Block-Based Programming Languages. In the first stage, on Machine Learning, the data set and the methods used in educational field were researched. In the second stage, the devices which are used to measure the performance on Block- Based Programming Languages were analyzed. On literature, there are various sources for teaching Computational Thinking, but it occurs that measuring instruments that evaluates Computational Thinking are inadequate. To evaluate Computational Thinking a machine learning model was trained. Trained model was turned into online evaluation tool by which both the tutors and the students evaluate Computational Thinking easily. In the final stage, the developed online evaluation tool and the outcome was analyzed. © 2020 IEEE.