Cancer is a life-threatening disease with high mortality rates. In the Indian subcontinent, women have a higher possibility to be diagnosed with cancer than men. The most common cancers identified in Indian women are Breast Cancer and Cervical Cancer. Both these cancers have high survival rates in case of early prediction. This paper reviews the attributes which are used in the existing datasets for prediction of these two cancers. The paper also proposes new attributes to overcome the limitations of existing ones, which will further increase the effectiveness of cancer prediction systems. The efficiency of existing and proposed attributes is compared by processing datasets through data mining algorithms using WEKA tool. The algorithms used for this study are – J48 (Decision Tree), Na?ve Bayes, Random Forest, Random Tree, KStar and Bagging Algorithm. The empirical analysis done in the paper reported improvement in the efficiency of cancer prediction over existing prediction systems.