Effect of Hyperparameter Tuning Using Random Search on Tree-Based Classification Algorithm for Software Defect Prediction
Year: 2024
Type: article
Abstract: The field of information technology requires software, which has significant issues. Quality and reliability improvement needs damage prediction. Tree-based algorithms like Random Forest, Deep Forest, and Decision Tree offer potential in this domain. However, proper hyperparameter configuration is crucial for optimal outcomes. This study demonstrates the use of Random Search Hyperparameter Setting Technique to predict software defects, improving damage estimation accuracy. Using ReLink datasets, we found effective algorithm parameters for predicting software damage. Decision Tree, Random Forest, and Deep Forest achieved an average AUC of 0.73 with Random Search. Random Search outperformed other tree-based algorithms. The main contribution is the innovative Random Search hyperparameter tuning, particularly for Random Forest. Random Search has distinct advantages over other tree-based algorithms
DOI: 10.22146/ijccs.90437
Institution: Lambung Mangkurat University
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $13
OpenAlex ID: W4392777930