The Impact of Feature Importance Methods on the Interpretation of Defect Classifiers

Rajbahadur, Gopi Krishnan; Ansaldi, Gustavo; Kamei, Yasutaka; Hassan, Ahmed E.

doi:10.1109/tse.2021.3056941

Cited by 54 publications

(25 citation statements)

References 115 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To see whether alternative model explaining method can generate similar results to SHAP, we used the permutation feature importance (PFI) method to determine the gene importance values. PFI is a widely used, generic approach for most classifiers ( Rajbahadur et al., 2021 ). In brief, we randomly shuffled the expression values for each gene, while keeping the labels intact.…”

Section: Resultsmentioning

confidence: 99%

Deep learning explains the biology of branched glycans from single-cell sequencing data

2022

View full text Add to dashboard Cite

Section: Resultsmentioning

confidence: 99%

Deep learning explains the biology of branched glycans from single-cell sequencing data

2022

View full text Add to dashboard Cite

“…Therefore, we further performed a feature visualization analysis, based on the XGBoost algorithm for feature importance analysis. This method has been extensively explored in existing studies [12][13][14]. Here, we follow the ideas of previous studies to explain the prediction results of the XGBoost model.…”

Section: Mae Mse Rmsementioning

confidence: 98%

High School English Performance Analysis Using Interpretable Machine Learning Approach

Qu¹,

Koay²

2022

Atlantis Highlights in Social Sciences, Education and Humanities

View full text Add to dashboard Cite

Currently, English learning is being taken to a new level for Chinese students. The transformation of English language teaching and learning is in full swing across the education sector in China. However, few studies have discussed the relationship between high school students' English language learning performance and other courses. To address these issues and gain a practical understanding of the relationship between students' English learning performance and other courses, this paper first collects learning data from 532 students in 10 courses at a high school in China. Second, this paper uses an integrated learning algorithm called XGBoost to predict the students' English learning performance. Specifically, the dataset is divided into a training set and a test set, and we train the model on the training set and test it on the test set. The test results show that the method in this paper can predict students' English performance well (MAE < 0.03, MSE < 0.07, RMSE < 0.04). Moreover, Chinese and mathematics scores were highly correlated with students' English scores. Based on the above findings, this paper further proposes relevant teaching suggestions. The results of this paper provide a practical reference for teaching on each campus.

show abstract

“…Step 5 Fig. 3 Reopened bug prediction model pipeline in prior studies (Jiarpakdee et al, 2019;Lee et al, 2020;Rajbahadur et al, 2021). Autospearman uses a criteria that selects one feature from a group of highest correlated features which shares the least correlation with other features that are not in the group (Jiarpakdee et al, 2018) based on Spearman correlation score.…”

Section: Feature Importancementioning

confidence: 99%

Revisiting reopened bugs in open source software systems

Tagra¹,

Zhang²,

Rajbahadur³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Reopened bugs can degrade the overall quality of a software system since they require unnecessary rework by developers. Moreover, reopened bugs also lead to a loss of trust in the end-users regarding the quality of the software. Thus, predicting bugs that might be reopened could be extremely helpful for software developers to avoid rework. Prior studies on reopened bug prediction focus only on three open source projects (i.e., Apache, Eclipse, and OpenOffice) to generate insights. We observe that one out of the three projects (i.e., Apache) has a data leak issue -the bug status of reopened was included as training data to predict reopened bugs. In addition, prior studies used an outdated prediction model pipeline (i.e., with old techniques for constructing a prediction model) to predict reopened bugs. Therefore, we revisit the reopened bugs study on a large scale dataset consisting of 47 projects tracked by JIRA using the modern techniques such as SMOTE, permutation importance together with 7 different machine learning models. We study the reopened bugs using a mixed methods approach (i.e., both quantitative and qualitative study). We find that: 1) After using an updated reopened bug prediction model pipeline, only 34% projects give an acceptable performance with AUC 0.7. 2) There are four major reasons for a bug getting reopened, that is, technical (i.e., patch/integration issues), documentation, human (i.e., due to incorrect bug assessment), and reasons not shown in the bug reports. 3) In

show abstract

The Impact of Feature Importance Methods on the Interpretation of Defect Classifiers

Cited by 54 publications

References 115 publications

Deep learning explains the biology of branched glycans from single-cell sequencing data

Deep learning explains the biology of branched glycans from single-cell sequencing data

High School English Performance Analysis Using Interpretable Machine Learning Approach

Revisiting reopened bugs in open source software systems

Contact Info

Product

Resources

About