This paper presents a highly effective technique to handle data insufficiency issue in software defect prevention using machine learning techniques. Extracting knowledge using data mining techniques in software engineering is a difficult task as the data available from software projects for research is not only less but also outdated. Generally Software engineering activities like defect prediction, effort estimation etc., were done on data available from open source datasets which is less in volume. All researchers and data scientists tend to agree on one common thing i.e. they always need more quality data to produce accurate results. When the data used to construct models is lesser than essential quantity, the results predicted will be inaccurate and unstable. In this paper, transfer learning technique has been employed to tackle this data insufficiency issue using techniques, by transferring knowledge from related similar task, where sufficient data is available and this extracted knowledge is made use in the pursuing task to create more accurate prediction. From the experimental results, it is evident that transfer learning technique employed show considerable improvement in defect prevention even when the data available for that problem is limited.

Transfer Learning Machine Learning Artificial Neural Networks