Software defect prediction ensemble learning algorithm based on adaptive variable sparrow search algorithm

Tang, Yu

doi:10.1007/s13042-022-01740-2

Software defect prediction ensemble learning algorithm based on adaptive variable sparrow search algorithm

2023-01-06

31

PRE

AI

DOI：10.1007/s13042-022-01740-2

Original

Original request for help

Save

Abstract

En 中文

Software defect prediction has caused widespread concern among software engineering researchers, which aims to erect a software defect prediction model according to historical data. Among all the techniques used in this field, extreme learning machine is widely used by researchers because of its simple structure and excellent learning speed. At the same time, the prediction performance of extreme learning machine is greatly affected by the random selection of parameters and the weak generalization ability. In this sense, in order to improve the prediction performance of the model, researchers uses swarm intelligence optimization algorithm to provide the optimal parameters for the model. Sparrow search algorithm is a new meta-heuristic algorithm that simulates the foraging and anti-predation behavior of the sparrow group. However, the original sparrow search algorithm is easily trapped to local optimal solutions in the later stage of the iterations. To improve the global optimization ability of the original sparrow search algorithm, this paper proposed an adaptive variable sparrow search algorithm (AVSSA) based on adaptive hyper-parameters and variable logarithmic spiral. This work run experiments of AVSSA in eight benchmark functions, and obtained the satisfactory results. In the traditional software defect prediction algorithm, the imbalance of data distribution is also one of the main reasons that affect the performance of the model. Therefore, this paper uses the adaptive variable sparrow search algorithm to optimize the extreme learning machine as the base predictor for Bagging ensemble learning (AVSEB). A new software defect prediction ensemble learning model is proposed in this paper. Firstly, the model used the unstable cut-points algorithm to preprocess Bagging sample set in this model. Then, the adaptive variable sparrow search algorithm is used to optimize the extreme learning machine as the base predictor of ensemble learning. Finally, the voting method is used to output the prediction results of software defects. Based on the experimental results, the evaluation index of our proposed algorithm is significantly superior to the other four advanced comparison algorithms in 15 open software defect datasets. According to the test results of Friedman ranking and Holm's post hoc test, this paper proposed algorithm has obvious statistical significance compared with other advanced prediction algorithms.

Keywords:

Sparrow search algorithm

Adaptive hyper-parameters

Variable logarithmic spiral

Extreme learning machine

Ensemble learning

Software defect prediction

Software defect prediction ensemble learning algorithm based on adaptive variable sparrow search algorithm

Abstract

Journal

Organization

Cited Papers