Hyperparameter Tuning for Overlapped Software Defect Prediction Data sets

Authors

  • Shivani Gupta , Kusum Lata Jain , Smaranika Mohapatra3

Abstract

The machine learning algorithms has become increasingly widely in many applications and
research. The classifiers in machine learning build a model from data that allows computers to
improve future predictions.
Despite the popularity of most machine learning algorithms require knowledge to take the decisions
about the appropriate model and parameter settings for a particular domain of problem. It’s very
difficult to identify the machine learning classifier that is most well suited for specific characteristics
of data.
In this work, we present a method to tune the hyperparameters of best machine learning algorithm
for a overlapped software defect data-set. The results obtained shows that using the machine learning
algorithm and hyperparameter tuning suggested by our method improves the predictive performance
of a classifier than its default settings on overlapped data-sets

Published

2020-11-01

Issue

Section

Articles