Data analytics and the novice programmer

Publication Type:
Thesis
Issue Date:
2018
Full metadata record
Files in This Item:
Filename Description Size
01front.pdf95.56 kB
Adobe PDF
02whole.pdf4.77 MB
Adobe PDF
The aptitude of students for learning how to program (henceforth Programming learn-ability) has always been of interest to the computer science education researcher. This issue of aptitude has been attacked by many researchers and as a result, different algorithms have been developed to quantify aptitude using different methods. Advances in online MOOC systems, automated grading systems, and programming environments with the capability of capturing data about how the novice programmer’s behaviour has resulted in a new stream of studying novice programmer, with a focus on data at large scale. This dissertation applies contemporary machine learning based analysis methods on such “big” data to investigate novice programmers, with a focus on novices at the early stages of their first semester. Throughout the thesis, I will demonstrate how machine learning techniques can be used to detect novices in need of assistance in the early stages of the semester. Based on the results presented in this dissertation, a new algorithm to profile novices coding aptitude is proposed and its’ performance is investigated. My dissertation expands the range of exploration by considering the element of context. I argue that the differential patterns recognized among different population of novices is very sensitive to variations in data, context and language; hence validating the necessity of context-independent methods of analyzing the data.
Please use this identifier to cite or link to this item: