Breast cancer is a heterogeneous and complex disease, with a high mortality rate. it is one of the predominant cancer types that affects millions of cases and causes thousands of deaths every year in india. the effective treatment and prognosis of breast cancer development relies largely on a correct classification and indeed may also help patients in avoiding improper chemotherapy and its side effects. in our studies, we developed a classifier to classify the five intrinsic subtypes of breast cancer using microarray gene expression profile data. the model is based on supervised learning using support vector machine algorithm for classification. the genes required for acting as the important features for classification were identified with a greedy algorithm i.e. recursive feature elimination method via logistic regression. we identified a panel of 40 genes which serve as the important features for classifying the breast cancer intrinsic subtypes, using their gene expression profile data. the modeled classifier has an accuracy score of 93-96% and it is more than the previous methods. it has good precision, recall and f score for prediction. this method uses minimal number of genes as features, which indicates that lesser the number of features, lesser will be the number of gene probes that can be used in diagnostic chips to detect breast cancer and its subtype. in the identified panel of genes, it is found that some of them are involved directly in the cancer process which may act as prognostic signatures.