Exploring the Behavioural Factors of Cervical Cancer Using ANOVA and Machine Learning Techniques
Science, Engineering Management and Information Technology First International Conference, SEMIT 2022, Virtual, Online, 2 - 03 February 2022, vol.1808 CCIS, pp.249-260, (Full Text)
- Publication Type: Conference Paper / Full Text
- Volume: 1808 CCIS
- Doi Number: 10.1007/978-3-031-40395-8_18
- City: Virtual, Online
- Page Numbers: pp.249-260
- Keywords: ANOVA F-Test, Cancer Prediction, Decision Tree, kNN Classification, Navie Bayes
- Ankara Yıldırım Beyazıt University Affiliated: Yes
Abstract
Cervical cancer is the fourth most common and one of the deadliest types of cancer in the female population. However, it does not show any symptoms in the first stage. Therefore, early diagnosis is very difficult. On the other hand, there are some risk factors directly or indirectly associated with it. The effect of these risk factors on predicting the diagnosis of cervical cancer was investigated with machine learning-based algorithms and promising results were obtained. However, studies on which risk factors are more effective than others are scarce. This study focuses on investigating how the selection of promising factors associated with cervical cancer will affect the predictive ability of machine learning-based classification algorithms such as Gaussian Naive Bayes (GNB), k Nearest Neighbor and Decision Tree. ANOVA F-test is applied to evaluate each risk factor independently according to the desired class. In order to ensure the reliability of the results, K-fold cross validation technique is used at different K values. In the selected cervical cancer behavioral risk dataset, the GNB algorithm showed the highest performance (%94) with eight risk factors for all K values.