A Framework for Software Defect Prediction Using Feature Selection and Ensemble Learning Techniques

Abstract

Persistent and quality graduation rates of students are increasingly important indicators of progressive and effective educational institutions. Timely analysis of students' data to guide instructors in the provision of academic interventions to students who are at risk of performing poorly in their courses or dropout is vital for academic achievement. In addition there is need for performance attributes relationship mining for the generation of comprehensible patterns. However, there is dearth in pieces of knowledge relating to predicting students' performance from patterns. This therefore paper adopts hierarchical cluster analysis (HCA) to analyze students' performance dataset for the discovery of optimal number of fail courses clusters and partitioning of the courses into groups, and association rule mining for the extraction of interesting course-status association. Agglomerative HCA with Ward's linkage method produced the best clustering structure (five clusters) with a coefficient of 92% and silhouette width 0.57. Apriori algorithm with support (0.5%), confidence (80%) and lift (1) thresholds were used in the extraction of rules with student's status as consequent. Out of the twenty one courses offered by students in the first year, seven courses frequently occur together as failed courses, and their impact on the respective students' performance status were assessed in the rules. It is conjectured that early intervention by the instructors and management of educational activities on these seven courses will increase the students' learning outcomes leading to increased graduation rate at minimum course duration, which is the overarching objective of higher educational institutions.

Software Defect Prediction Using Variant based Ensemble Learning and Feature Selection Techniques

“…These datasets are currently available at [24]. In this research we have used the DS'' version of NASA datasets which is already been used by many researchers [1,2,3,4,5,25,26,27]. Second stage of the framework deals with the selection of best variants from different classifiers (Fig.3).…”

“…In Data preprocessing, two tasks are performed: Resampling [31,32] and Randomization. Resampling is performed to resolve the issue of class imbalance in the datasets as this issue can compromise the accuracy of proposed classification framework [2,3,4,5]. To perform this task, the builtin function of WEKA is used (weka.filters.supervised.instance.Resample).…”

“…For classification, various widely used machine learning techniques are used and effects of the resampling techniques on the performance is analyzed by using four measures such as: "F-measure, Accuracy, MCC and ROC". Researchers in [4] presented a framework to predict the defects at early stages of software development. To increase the performance of prediction, the researchers incorporated feature selection and ensemble learning techniques and for performance evaluation four measures are used: F-measure, Accuracy, MCC and ROC.…”

“…The prediction of software defects is a binary classification problem as we have to identify that the particular module is defective or non-defective. Many researchers have focused on machine learning techniques to solve the binary classification problems such as: Network Intrusion Detection [21,22], Sentiment Analysis [13,14,15,16,17,18], Rainfall Prediction [19,20], and Software Defect Prediction [1,2,3,4,5]. The process of software defect prediction has been focused by many researchers in the last decade however improving the prediction accuracy has always been the main concern.…”

Prediction of Defect Prone Software Modules using MLP based Ensemble Techniques

“…The results reflected that "Random Over Sampling" performed well among other techniques. Researchers in [4] presented a feature selection based ensemble classification framework to predict defect prone software modules. The framework is implemented on six publically available Cleaned NASA MDP datasets and the performance is analyzed by using F-measure, Accuracy, MCC and ROC.…”