Why even my training data failed during the prediction of libsvm

Question

Currently I'm using libsvm for my one class classification problem. I have 10 samples in my training set, 5 samples in my testing set, both of my training and testing set is scaled by svm_scale, then I use smv_train with the scaled training set for training, and svm_pridict with the scaled testing set for testing. But this problem occurs which makes me very confusing. In the result, why even my training sample fails....

Accuracy = 26.666666667% (4/15) (classification)

Answer 1

It's likely you need more data. It's very hard for any algorithm to generalize well on 10 samples, particularly an SVM that is trying to find the subset of data points (i.e. less than 10) that best segregate the data.