EPFSESD: An Enhanced Pipeline Feature Selection Algorithm for Erythemato-Squamous Disease Detection

  • Rajashekar Deva, Dr. G .Narsimha
Keywords: Enhanced Pipelined Feature Selection, Random Forest, Recursive Elimination Feature

Abstract

Erythemato-Squamous Disease (ESD) is one of the common skin diseases. In machine learning, the current trends are focusing on automation of health care, agriculture, and manufacturing domains. The proposed system automates the Erythemato-Squamous Disease (ESD) to help the skin doctors for identification of skin diseases easily. In this paper, the system utilizes the dataset that is freely available at the UCI repository, which contains 34 features, which takes high computation time for determining the disease. To overcome this problem, the proposed system performs dimensionality reduction by identifying the important features and the selection of important attributes plays a vital role in the detection of class labels of the disease. The dataset contains 6 classes, multi-class problem. In Multi-class problems, the best features are identified using an enhanced pipeline algorithm in combination with recursive elimination feature, an embedded method to design a robust system. The pipeline algorithm has sequenced 5   different types of classifiers Logistic Regression, Multi-Layer Perceptron, CART, Random Forest, and Gradient Boost Algorithm.

Published
2021-06-29
How to Cite
Dr. G .Narsimha, R. D. (2021). EPFSESD: An Enhanced Pipeline Feature Selection Algorithm for Erythemato-Squamous Disease Detection. Design Engineering, 631-646. Retrieved from http://thedesignengineering.com/index.php/DE/article/view/2323
Section
Articles