원문정보
초록
영어
Ovarian cancer is very malignant tumor because it doesn’t have any striking symptoms in its early stages. That’s why the early screening is really necessary in its clinics. We try to look for the optimal methodology to find out biomarker combination making its classification performance better than other cases. We evaluate 9 machine learning algorithms, those are Random Forest, Logistic, Multilayer Perceptron, Bagging, Classification Via Regression, LogitBoost, MultiClassifer, Simple Logistic, and Logistic Regression. The Area Under the Curve (AUC) of each algorithm is compared. We firstly select 15 biomarkers which are widely spread in the ovarian cancer diagnosis and find the best three combinations which composed of two, three and four biomarkers by using Logistic Regression which is well known for its reliable performance. Than we re-evaluate the best combinations with nine algorithms including Logistic Regression to find the optimal machine learning algorithm. In this research, we can find possibility to use another machine learning algorithm rather than Logistic Regression.
목차
1. Introduction
2. Data Set
3. Experiment
4. Results
5. Conclusion
Acknowledgements
References