K-means based quality prediction of object-oriented software using LR-ACO

Abstract

A quality prediction mechanism has been developed in this paper. K-means clustering algorithm has been applied for the clustering of object-oriented features. Finally logistic regression (LR) and ant colony optimization (ACO) (LR-ACO) have been used for the classification. The object-oriented parameters have been considered like polymorphism, encapsulation, abstraction, inheritance and other object-oriented features for experimentation. The purpose of these features to categorize the data in different class levels based on memory usage, reusability and multiple forms. Different hyperparameters like dynamic allocation and feature margin have also been considered for the classification thresholds. Different performance measures have been considered for the experimentation and the results shows the approach effectiveness through different exploration.