³ÉÈËÖ±²¥app

Research area(s)

• Software engineering,
• Empirical research,
• Cost modelling and prediction,
• Machine learning (including case-based reasoning, metaheuristics, rule induction algorithms and Grey relational algebra),
• Data imputation and noise handling,
• Reproducibility, replicability and meta-analysis.

Project details

Presently I’m interested in the impact of bias amongst computer scientists when conducting and reporting computational experiments. A meta-analysis conducted with Tracy Hall and David Bowes (Univ. of Hertfordshire) of 600 experimental results from studies exploring how we can induce classifiers to predict whether software will be faulty or not shows that who does the work is 25x more influential than what algorithms are deployed. This was published in IEEE Transactions on Software Engineering [1] and has more than 140 citations to date.

I’m also working on a project with cognitive psychologists to experimentally investigate the various biases software professionals are vulnerable to when making predictions. Most noteworthy is the impact of the anchoring bias [2].

[1] M. Shepperd, D. Bowes, and T. Hall, “Researcher Bias: The Use of Machine Learning in Software Defect Prediction,” IEEE Transactions on Software Engineering, vol. 40, no. 6, pp. 603-616, 2014.

[2] M. Shepperd, C. Mair, and M. Jørgensen, “An Experimental Evaluation of a De-biasing Intervention for Professional Software Developers,” in 33rd ACM Software Applications Conference (SAC’18), Pau, France, 2018.