I've got 700 data rows with about 6 features and there are 300 different class labels. Most of the classes have got only 2 data rows (ca. 250 classes).
I know there are too many classes for this data set. But I have to do some kind of classification with Machine Learning (scikit).
I already tried SVM, but the result was devastating. My prediction rate for a class was under 1%.
Is there an approach to get a better prediction rate with this data set?