Broadly speaking, his research focuses on software engineering and data science. The core of his research is with respect to:
- Predictive Analytics for Software Engineering | Exploring new technologies to ensure highly accurate prediction models
Predictive analytics play a big role in software engineering research. For example, it has used to predict software defects and understand their characteristics for several decades.
His research aims to explore new techniques to ensure highly accurate and reliable predictions, and insights derived from prediction models. In particular, he investigate the impact that experimental components of defect prediction modelling have on the predictions and insights.
For example, his research shows that (1) noise in defect datasets, (2) the choice of parameter settings of classification techniques, and (3) the choice of model validation has a large impact on the performance and interpretation of defect prediction models. Moreover, he also shows that collinearity and multicollinearity have a large impact on the stability of the insights derived from prediction models.
- New Technology for Software Development Process | Improving quality and efficiency throughout the software-development process
In modern software development, software systems continue to grow in size and complexity due to the increasing of software requirements. For example, the Eclipse software that was released in 2014 has more than 2 million lines of code. This explosive growth of software size has increased more rapidly than the ability that human can maintain them. Such maintenance activities involve with various key activities, such as, software debugging, peer code review. His research aims to develop new technologies to improve software developers' productivity throughout the software-development process. For example, he proposed approaches to (1) recommend developers to review new source code changes before integrating them to a codebase; and (2) recommend source codes that be fixed for a given bug report.
- Social Analytics | Investigating user behaviors
With an increasing amount of data on every aspect of our daily activities - from what we buy, to where we travel, to who we know, and beyond - we are able to measure human behavior and investigate interesting correlations which can be used to predict users behaviors, preferences or characteristics. For example, he investigates patterns of Internet usage through millions web traffic logs of Kasetsart University, Thailand, in order to better understand (1) the most popular websites, (2) traffic workload, (3) time spent on the Internet, and (4) web access pattern.