Paper Info
Title
Using unstructured practice plus reflection to develop programming/problem-solving fluency.
Abstract
This paper describes a problem-solving course for second-year undergraduate students, designed to foster programming fluency, and explores the factors associated with the acquisition of that fluency. The course introduces algorithmic techniques such as brute force, recursion, dynamic programming and graph algorithms. The course structure revolves around repeated practice: students are required to apply programming skills using many small but challenging coding problems, combined with reflection reports and journal writing to document their thoughts and guide their design. Progress is assessed by three practical examinations where students solve a graduated series of small open-ended programming challenges under time constraints. Challenges of this type are often used in technical interviews in industry and can be viewed as one indicator of programming fluency and problem-solving ability. However, while there is a progression in the complexity of problems that students are able to attempt during the course, it is not clear which factors are the strongest contributors to students' performance at practical exams. This work uses a large ensemble of symbolic regression models to identify factors that related strongly to problem-solving performance. Through this stochastic analysis we are able to identify which factors are most consistently predictive of exam performance. To help validate these findings we conduct a broad analysis of coded reflections. It is hoped these findings will help identify pre-cursor skills and behaviours that can be the target of future interventions.
Year
DOI
Venue
2018
10.1145/3160489.3160496
ACE 2018: 20th Australasian Computing Education Conference Brisbane Queensland Australia January, 2018
Field
DocType
ISBN
Dynamic programming,Psychological intervention,Fluency,Computer science,Stochastic process,Coding (social sciences),Brute force,Artificial intelligence,Symbolic regression,Recursion,Machine learning
Conference
978-1-4503-6340-2
Citations 
PageRank 
References 
0
0.34
6
Authors
2
Name
Order
Citations
PageRank
Cruz Izu114923.41
Bradley Alexander211312.67