Decision Tree based Classification and Dimensionality Reduction of Cervical Cancer

Abstract

The data revolution in medicines and biology have increased our fundamental understandings of biological processes and determining the factors causing any disease, but it has also posed a challenge towards their analysis. After breast cancer, most of the deaths among women are due to cervical cancer. According to IARC, alone in 2012 a noticeable number of cases estimated 7095 of cervical cancer were reported. 16.5% of the deaths were due to the cervical cancer with the total deaths of 28,711 among women. To analyze the high dimensional data with high accuracy and in less amount of time, their dimensionality needs to be reduced to remove irrelevant features. The classification is performed using the recent iteration in Quinlan’s C4.5 decision tree algorithm i.e. C5.0 algorithm and PCA as Dimensionality Reduction technique. Our proposed methodology has shown a significant improvement in the account of time taken by both algorithms. This shows that C5.0 algorithm is superior to C4.5 algorithm.