Comparative analysis on prediction models with various data preprocessings in the prognosis of cervical cancer

Abstract

Cancer is a life-threatening disease, which is considered incurable most of the time. Hence, an early prediction for a possible risk of cancer can be very helpful in order to start treatment as soon as possible. One such cancer is Cervical Cancer. Here, we have used data about the lifestyle and previous medical history of women in order to try and predict if a woman is susceptible to cervical cancer or not. However, the dataset needed a lot of preprocessing for handling missing values. Four different techniques were used in order to fill in the missing values. Soon after, the correlation among the variables was calculated to keep the most effective ones. Then, prediction algorithms namely Decision Tree, Random Forest, Logistic Regression, Naïve Bayes, Support Vector Machine and Neural Network were used to train and test for positive and negative diagnosis. Our findings were that SVM and Logistic Regression had the highest Precision, Recall, F1 Score and Accuracy.