KDV classifier: a novel approach for binary classification

The current era is an era of Artificial Intelligence. Artificial intelligence is an umbrella discipline that includes Machine Learning as a crucial component. In the Machine Learning space, Classification is an important research area that cannot be neglected. We can define classification as systematically arranging objects or elements in different groups based on given conditions or criteria. An important class of classifier is a Binary classifier that classifies observations or data into two classes. The binary classifier is useful when observation can only be grouped in two categories or where classification in two classes is required in a given situation. One example of a binary classifier is whether a patient is cancerous or not. In literature many binary classification algorithms are available. The proposed classifier in this research paper is also a binary classifier. The name of the proposed classifier is KDV Binary Classifier. Here KDV stands for K-Distance Variance. K-Distance is the distance of the kth nearest object of a given data point. This binary classifier is particularly useful if observations are not balanced. One particular class outnumbers another class. We compared KDV with KNN for binary classification based on the percentage of accuracy. KNN is a general classifier. We considered its binary aspect. The result shows that KDV is comparable with KNN. Many times KDV outperforms KNN. We compared results for accuracy using cross-validation methods like twofold, fivefold, tenfold and the Also Leave one out method. KDV can be a good research area in the field of Machine Learning.

Automated summary

This article introduces the important research area of classification in machine learning, particularly binary classification. It presents a binary classifier called KDV, which is compared to the general classifier KNN to demonstrate its advantages. The article also examines the accuracy of KDV using various cross-validation methods, and suggests that KDV has research potential in the field of machine learning.