Feature Selection Using Firefly Algorithm With Tree-Based Classification In Software Defect Prediction

Defects occurring in software product are a universal event. Prevention of these defects in the early stage needs more attention because early stage prevention and fixing requires less effort and lower cost. Software defect prediction (SDP) is necessary in the determination of software quality as well as reliability. Prediction of defects is relatively an original research area in software quality engineering. Coverage of key predictors and the kind of data to be collected along with defect prediction model role, the interdependence of defects and predictors can be recognized in software quality. Feature selection (FS) is one of the worthy preprocessing techniques for application that uses huge volumes of data. It is the process of selecting the probable minimal attribute which is expected to be represented in the set of actual attributes. This paper proposes, FS using firefly algorithm (FA) and classifiers like support vector machine (SVM), Naive Bayes (NB) as well as K-nearest neighbor (KNN) are used for classifying the features selected. The FS that make use of the FA is that new technique of evolutionary computation that has been inspired by the process of flash lighting of the fireflies. This can search quickly the feature space for an optimal or a near optimal feature subset for minimizing a certain function of fitness. This proposed fitness function has made use of the incorporation of both the accuracy of classification and the reduction of the size. The results of the experiment have shown that the FS using the FA can achieve a better accuracy of classification than that of the other methods.

Applying change‐level software defect prediction (SDP) in practice has several challenges regarding model validation techniques, data accuracy, and prediction performance consistency. A few studies report on these challenges in an industrial context. We share our experience in integrating an SDP into an industrial context. We investigate whether an “offline” SDP could reflect its “online” (real‐life) performance, and other deployment decisions: the model re‐training process and update period. We employ an online prediction strategy by considering the actual labels of training commits at the time of prediction and compare its performance against an offline prediction. We empirically assess the online SDP's performance with various lengths of the time gap between the train and test set and model update periods. Our online SDP's performance could successfully reach its offline performance. The time gap between the train and test commits, and model update period significantly impacts the online performance by 37% and 18% in terms of probability of detection (pd), respectively. We deploy the best SDP solution (73% pd) with an 8‐month time gap and a 3‐day update period. Contextual factors may determine the model performance in practice, its consistency, and trustworthiness. As future work, we plan to investigate the reasons for fluctuations in model performance over time.

Software defect prediction enhances the quality, efficiency, and effectiveness of time and expenses for software testing by focusing on defect modules. Software defect prediction technology uses machine learning to predict defect modules, making allocating limited resources easier quickly. Software defect prediction datasets naturally have imbalanced class problems with significantly few defective modules compared to non-defective modules. In this study, software defect prediction data was classified by implementing the LSSVM algorithm with ReliefF (K=10) feature selection and applying the SMOTE method to overcome the imbalanced class problem in the dataset. The datasets used are software defect prediction datasets of the public NASA MDP Promise project, namely CM1, PC1, KC1, and KC2. Dataset divided into training and testing data using Fold Cross-Validation with ten folds. The classifier achieved the highest average accuracy on the PC1 dataset, which was 93,87%, while the highest Area Under the ROC Curve (AUC) was achieved by the classifier for the KC2 dataset, which was 78,35%. The results also indicate that AUC values for classifiers that use SMOTE always higher than non-SMOTE in each dataset

Software defect prediction (SDP) is used to identify defects in software modules that can be a challenge in software development. This research focuses on the problems that occur in Particle Swarm Optimization (PSO), such as the problem of noisy attributes, high-dimensional data, and premature convergence. So this research focuses on improving PSO performance by using feature selection methods with hybrid techniques to overcome these problems. The feature selection techniques used are Filter and Wrapper. The methods used are Chi-Square (CS), Correlation-Based Feature Selection (CFS), and Forward Selection (FS) because feature selection methods have been proven to overcome data dimensionality problems and eliminate noisy attributes. Feature selection is often used by some researchers to overcome these problems, because these methods have an important function in the process of reducing data dimensions and eliminating uncorrelated attributes that can cause noisy. Naive Bayes algorithm is used to support the process of determining the most optimal class. Performance evaluation will use AUC with an alpha value of 0.050. This hybrid feature selection technique brings significant improvement to PSO performance with a much lower AUC value of 0.00342. Comparison of the significance of AUC with other combinations shows the value of FS PSO of 0.02535, CFS FS PSO of 0.00180, and CS FS PSO of 0.01186. The method in this study contributes to improving PSO in the SDP domain by significantly increasing the AUC value. Therefore, this study highlights the potential of feature selection with hybrid techniques to improve PSO performance in SDP.

Software defects and quality assurance are crucial aspects of software development that should be considered during the software development cycle. To ensure high-quality software, it is essential to have a robust quality assurance process in place. System reliability and quality are very key components that must be considered during software development, and this can only be achieved when software undergoes a thorough test process for errors, anomalies, defects, omissions, and bugs. Early software defect prediction and detection play an essential role in ensuring the reliability and quality of software systems, ensuring that software companies discover errors or defects early enough and allocate more resources to defect-prone modules. This study proposes the development of an enhanced classifier model for software defect prediction and detection. The aim is to harness the collective intelligence of selected base classifiers like Support Vector Machine, Logistic regression, Decision Trees, Random Forest, AdaBoost, Gradient Boosting, K-Nearest Neighbor, GaussianNB, and Multi-Layer Perception to improve accuracy, robustness, and generalization in identifying potential defects using a soft voting ensemble technique. The ensemble model leveraged the confidence probability of the soft voting technique and the generalization advantage of cross-validation leading to a more robust and dynamic model. The performance of the model with existing classifiers was evaluated using accuracy, F1 score, Precision, and area under the ROC curve (ROC- AUC) as the evaluation metrics. The results of the experiment revealed that the Proposed Classifier produced an overall Accuracy rate of 93%, and ROC AUC of 98%. The results demonstrate the effectiveness of our enhanced ensemble classifier in software defect detection and prediction. By harnessing the strengths of diverse base classifiers, our approach provides a robust and adaptive solution to the challenges of early detection and mitigating defects in software systems. This research contributes to the advancement of reliable software development practices and lays the foundation for future enhancements in ensemble-based defect detection methodologies.

It is known that certain classification algorithm requires continuous data to be discretized for it to produce better classification accuracy. Hence, many works have explored the pairing of classification algorithm and discretization techniques, yet tree-based classifier especially Classification and Regression Trees (CART) still have an issue with classification accuracy regardless of different pairing with existing discretization techniques. The role of fuzzy partition and fuzzy sets interval are not something new in data discretization but none yet to explore the pairing of fuzzy discretization with tree-based algorithm. This paper will be discussing on an approach of using fuzzy based discretization and a member of tree-based algorithm known as Random Forest, a better version of CART. In this study, continuous data are identified from a dataset and discretized through the fuzzy discretization. Then, 10-fold cross validation is done on the transformed dataset and seven well-known classifiers are used including the proposed approach. Based on the results, better classification accuracy is achieved when fuzzy discretization is paired with Random Forest algorithm compared to CART. On top of that, with the present of fuzzy discretization technique, an increased in the classification accuracy has been obtained compared to other classification algorithms.

Finding defects early in a software system is a crucial task, as it creates adequate time for fixing such defects using available resources. Strategies such as symmetric testing have proven useful; however, its inability in differentiating incorrect implementations from correct ones is a drawback. Software defect prediction (SDP) is another feasible method that can be used for detecting defects early. Additionally, high dimensionality, a data quality problem, has a detrimental effect on the predictive capability of SDP models. Feature selection (FS) has been used as a feasible solution for solving the high dimensionality issue in SDP. According to current literature, the two basic forms of FS approaches are filter-based feature selection (FFS) and wrapper-based feature selection (WFS). Between the two, WFS approaches have been deemed to be superior. However, WFS methods have a high computational cost due to the unknown number of executions available for feature subset search, evaluation, and selection. This characteristic of WFS often leads to overfitting of classifier models due to its easy trapping in local maxima. The trapping of the WFS subset evaluator in local maxima can be overcome by using an effective search method in the evaluator process. Hence, this study proposes an enhanced WFS method that dynamically and iteratively selects features. The proposed enhanced WFS (EWFS) method is based on incrementally selecting features while considering previously selected features in its search space. The novelty of EWFS is based on the enhancement of the subset evaluation process of WFS methods by deploying a dynamic re-ranking strategy that iteratively selects germane features with a low subset evaluation cycle while not compromising the prediction performance of the ensuing model. For evaluation, EWFS was deployed with Decision Tree (DT) and Naïve Bayes classifiers on software defect datasets with varying granularities. The experimental findings revealed that EWFS outperformed existing metaheuristics and sequential search-based WFS approaches established in this work. Additionally, EWFS selected fewer features with less computational time as compared with existing metaheuristics and sequential search-based WFS methods.

Role of risk factors and their variable types in predicting noise-induced hearing loss using artificial intelligence algorithms.

The class overlap problem refers to instances from different categories heavily overlapping in the feature space. This issue is one of the challenges in improving the performance of software defect prediction (SDP). Currently, the studies on the impact of class overlap on SDP mainly focused on within-project defect prediction and cross-project defect prediction. Moreover, the existing class overlap instances cleaning methods are not suitable for cross-version defect prediction. In this paper, we propose a class overlap instances cleaning method based on the Ratio of K-nearest neighbors with the Same Label (RKSL). This method removes instances with the abnormal neighbor ratio in the training set. Based on the RKSL method, we investigate the impact of class overlap on the performance and interpretability of the cross-version defect prediction model. The experiment results show that class overlap can affect the performance of cross-version defect prediction models significantly. The RKSL method can handle the class overlap problem in defect datasets, but it may impact the interpretability of models. Through the analysis of feature changes, we consider that class overlap instances cleaning can assist models in identifying more important features.

Nowadays, software is very influential on various sectors of life, both to solve business needs, as well as personal needs. To have a Software with high quality, testing is needed to avoid software defect. Research on software defects involving Machine Learning is currently being carried out by many researchers. This method contains one important step, which is called feature selection. In this study, researchers conducted a feature selection based on the software metric category to determine the level of accuracy of the prediction of software defects by utilizing 13 (thirteen) datasets from NASA MDP namely CM1, JM1, KC1, KC3, KC4, MC1, MC2, MW1, PC1, PC2, PC3, PC4, and PC5. To classify, the researchers involved 5 (five) classifiers, namely Naive Bayes, Decision Trees, Random Forests, K-Nearest Neighbor, and Support Vector Machines. The research result shows that each attribure on software metric categories has effect on each dataset. Naive Bayes Algorithm and Random Forest Algorithm can give better performance than other algorithm in classifieng software defect with feature selection based on metrics. On the other hand, the best metrics category on each classifier algorithm is metric Misc. From average AUC value, it can be concluded that metrics category which can give best performance is metric LoC, followed by metric Misc. Both categories have achieved highest AUC value in Random Forest classifier.

Feature selection is one of the most significant procedures in machine learning algorithms. It is particularly to improve the performance and prediction accuracy for complex data classification. This paper discusses a hybrid feature selection technique with the decision tree-based classification algorithm. The feature selected using information gain (IG) is combined with the features selected from ReliefF which generates the resultant feature subset. Then the resultant feature subset is in turn combined with a correlation-based feature selection (CFS) method to generate the aggregated feature subset. To perform classification accuracy on the aggregated feature subset, different decision trees-based classification algorithm such as C4.5, decision stumps, naive Bayes tree, and random forest with ten-fold cross-validation have been deployed. To check the prediction accuracy of the proposed work eight different multilevel University of California, Irvine (UCI) machine learning datasets have been used with minimum to maximum numbers of features. The main objective of the hybrid feature selection is to improve the classification accuracy, prediction and to reduce the execution time using standard datasets.

The ability to detect land use and land cover change quickly and accurately is crucial for earth system modeling, policy making, and sustainable land management. Remote sensing has been widely used to map and monitor land use and land cover change over very large areas. Many change detection algorithms (CDAs) have been developed with promising accuracy. However, accuracy of detecting specific types of change using these algorithms is often not satisfactory owing to errors of commission. We present a novel pixel-based broad area search (BAS) approach that detects and classifies heavy construction, which is an important indicator of human development and of interest to the intelligence community. The BAS system combines an online CDA, roboBayes, with a supervised tree-based classifier that removes the CDA’s errors of commission. To assess the performance of the classifier, we examined three tree-based algorithms – decision tree, random forest, and LightGBM – trained on roboBayes model parameters, tuning the models using a leave-one-region-out cross-validation strategy. We compared the performance of the tree-based classifiers against a baseline of filters created by the authors. Performance was evaluated at the pixel-level using precision, recall, and F1-score, which are analogues of commission error, omission error, and accuracy, respectively. The BAS system with optimized tree-based filters performed nearly 80% better than the BAS system without any filters and more than 50% better than the authors’ filters.

Current decision trees such as C4.5 and CART are widely used in different fields due to their simplicity, accuracy and intuitive interpretation. Similar to other popular classifiers, these tree-based classification algorithms are developed for fixed-length vector data and suffer from intrinsic limitations in handling complex data such as sequences. To tackle the discrete sequence classification task, the dominant strategy is to adopt a two-step procedure: first transform the sequential dataset into a vector dataset and then apply existing tree-based classifiers on the new vector data. However, such methods are highly dependent on the feature generation procedure and some features that are critical to the tree construction may be missed. To alleviate these issues, we present a new tree-based sequence classification method, which is able to construct a concise decision tree from the feature space that is composed of all subsequences present in the training sequences. Experimental results on fourteen real datasets show that our method can achieve better performance than those state-of-the-art sequence classification algorithms. The source codes of our method are available at: https://github.com/ZiyaoWu/SeqDT.

Athletes are rigorously trained and practice for gradually improving their fitness, sports-specific skills, and motor qualities to achieve success in various competitions. Periodic testing of physiological and physical fitness components is done to assess the progress of athlete in the training. Athletes are also selected for the upcoming competitions using these measurement results. However, there is a lack of an objective method for evaluating and selecting athletes for competitions. Machine learning predictive models are developed for important feature selection and prediction of weightlifting performances at the national or international level and how much total weight an athlete could lift in the future competitions. Predictive modeling requires historical data; hence, five-year fitness testing and actual competition performances were used for the prediction. Various fitness components are often highly correlated, and multicollinearity is a major issue in developing regression models for predicting performances. Hence, regularized ridge, lasso, and elastic net regression and tree-based random forest algorithms were developed to predict the performances along with the conventional statistical regression models. Boruta algorithm was applied to identify the most important predictors of the weightlifting performances. Back and handgrip strength, lower body average power, upper body explosive power and strength, age, and height were the most important predictors of the weightlifting performance. Thus, random forest with confirmed most important variables is the best model to predict the total weight-lifted and the national/international performances in selecting the weightlifters objectively for the upcoming competitions.

Software defect prediction (SDP) is designed to assist software testing, which can reasonably allocate test resources to reduce costs and improve development efficiency. In order to improve the prediction performance, researchers have designed many defect-related features for SDP. However, feature redundancy (FR) and irrelevance caused by the increasing dimensions of data will greatly degrade the performance of defect prediction. In order to solve the problems, researchers have proposed various data dimensionality reduction methods. These methods can be simply divided into two categories of methods: feature selection and feature extraction. However, the two categories of methods have their own advantages and limitation. In this paper, we propose a Hybrid Feature Dimensionality Reduction Approach (HFDRA) for SDP, which combines the two different kinds of methods, to improve the performance of SDP. HFDRA approach can be divided into two stages: feature selection and feature extraction. First, HFDRA divides the original features into several feature subsets through a clustering algorithm in the feature selection stage. Then, in the feature extraction stage, kernel principal component analysis (KPCA) is used to reduce the dimensionality of each feature subset. Finally, the reduced-dimensional data is used to build the prediction model. In the empirical study, we use 22 projects from AEEEM, SOFTLAB, MORP, and ReLink as experiment object. In this paper, we first compare our approach with seven baseline methods and three state-of-the-art methods. Then, we analyze the relationship between FR and prediction performance. Experiment results show that our approach outperforms the state-of-the-art data dimensionality reduction methods for defect prediction.