Unfortunately, each of us has a 1/4 chance of getting cancer. Although with advances in treatment technologies, the survival rate of cancer patients has increased, cancer still kills many people. Breast cancer is the second most diagnosed cancer and most fatal in women. The goal of this project is to develop models that can accurately classify breast cancer patient outcomes as either "alive" or "dead", based on demographic data and clinical data at the time of diagnosis.
Data: The Cancer Genome Atlas Breast Cancer (TCGA-BRCA) project through the National Cancer Institute - GDC Data Portal.
Method: We extract patient clinical information of the patients and engineer the features as necessary. Then we apply a few classification algorithms such as random forest, AdaBoost, SVC, logistic regression, K-nearest neighbor, and MLP while keeping the decision tree algorithm as our base model to predict patients' vital status.