Code smells are characteristics of the software that indicate a code or design problem which can make software hard to understand, evolve, and maintain. The code smell detection tools proposed in the literature produce different results, as smells are informally defined or are subjective in nature. To address the issue of tool subjectivity, machine learning techniques have been proposed which can learn and distinguish the characteristics of smelly and non-smelly source code elements (classes or methods). However, the existing machine learning techniques can only detect a single type of smell in the code element which does not correspond to a real world scenario. In this paper, we have used multilabel classification methods to detect whether the given code element is affected with multiple smells or not. We have considered two code smell datasets for this work and converted them into a multilabel dataset.In our experimentation, Two multilabel methods performed on the converted dataset which demonstrates good performances in the 10-fold cross-validation, using ten repetitions.

No abstract

Code smell is an inherent property of software that results in design problems which makes the software hard to extend, understand, and maintain. In the literature, several tools are used to detect code smell that are informally defined or subjective in nature due to varying results of the code smell. To resolve this, machine leaning (ML) techniques are proposed and learn to distinguish the characteristics of smelly and non-smelly code elements (classes or methods). However, the dataset constructed by the ML techniques are based on the tools and manually validated code smell samples. In this article, instead of using tools and manual validation, the authors considered detection rules for identifying the smell then applied unsupervised learning for validation to construct two smell datasets. Then, applied classification algorithms are used on the datasets to detect the code smells. The researchers found that all algorithms have achieved high performance in terms of accuracy, F-measure and area under ROC, yet the tree-based classifiers are performing better than other classifiers.

No abstract

Code smell is an inherent property of software that results in design problems which makes the software hard to extend, understand, and maintain. In the literature, several tools are used to detect code smell that are informally defined or subjective in nature due to varying results of the code smell. To resolve this, machine leaning (ML) techniques are proposed and learn to distinguish the characteristics of smelly and non-smelly code elements (classes or methods). However, the dataset constructed by the ML techniques are based on the tools and manually validated code smell samples. In this article, instead of using tools and manual validation, the authors considered detection rules for identifying the smell then applied unsupervised learning for validation to construct two smell datasets. Then, applied classification algorithms are used on the datasets to detect the code smells. The researchers found that all algorithms have achieved high performance in terms of accuracy, F-measure and area under ROC, yet the tree-based classifiers are performing better than other classifiers.