Parameter Tuning of a Sampling Technique for Change Prediction
Ankita Bansal1, Abha Jain2
1Ankita Bansal, Department of Information Technology, Netaji Subhas University of Technology, Delhi, India.
2Abha Jain, Department of Computer Science, Shaheed Rajguru College of Applied Sciences for Women, University of Delhi, India.
Manuscript received on 30 September 2019 | Revised Manuscript received on 12 November 2019 | Manuscript Published on 22 November 2019 | PP: 1718-1722 | Volume-8 Issue-6S3 September 2019 | Retrieval Number: F13220986S319/19©BEIESP | DOI: 10.35940/ijeat.F1322.0986S319
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Change prediction is very essential for producing good quality software. It leads to saving of lots of resources in terms of money, manpower and time. Predicting the classes during early phases can be done with the help of model construction using machine learning techniques. Every technique requires approximately equal distribution of classes (balanced data) for an efficient prediction. In this study, we have used a sampling approach to balance the data. We observed the improvement in accuracy after the models are trained on the balanced data. To further improve the accuracy of the models, the default parameters of the sampling approach have been adjusted /tuned. The results show the improvement in accuracy after sampling and parameter tuning.
Keywords: Class Imbalance, Machine Learning, Metrics, Quality, Resampling, Sampling, Tuning.
Scope of the Article: Regression and Prediction