Gender prediction of Indian names

We present a Support Vector Machine (SVM) based classification approach for gender prediction of Indian names.We first identify various features based upon morphological analysis that can be useful for such classification and evaluate them. We then state a novel approach of using n-gram-suffixes along with these features which gives us significant advantage over the baseline approach. We believe that we are the first to use n-grams of suffixes instead of the whole word for predictor systems. Our system reports a top F₁ score of 94.9% which is expected to improve further with increase in training data size.