DefinePK

Abstract

Lung cancer is one of the most dazedly disease around the world. It is not only affecting the people who are active smokers, but this disease is also affecting the people who are not smoking. Therefore, it is essential to investigate, which features are most responsible for lung cancer. Therefore, the proposed work is providing the two main contributions: (1) providing the detailed understanding about the different essential attributes, which are highly responsible for lung cancer risk. (2) Providing a machine learning model for accurately predicting the lung cancer risk. In this context, first a publically available dataset has been considered for study. Additionally, the feature relevance analysis has been performed using the random forest classifier. Further, by using a fixed threshold (0.01) the less relevant features are eliminated. Further, each selected attribute has been discussed and their details are provided. Finally two classifiers namely decision tree and convolutional Neural network (CNN) has been trained and validated. The validation results provide 100% accuracy for both the machine learning algorithm. Therefore, the prediction rules have been prepared for accurately predict that can predict the lung cancer level accurately.