Xtra-Computing / thundersvm

ThunderSVM: A Fast SVM Library on GPUs and CPUs
Apache License 2.0
1.55k stars 215 forks source link

question about the functions of predict and predict_proba #235

Open zhDai opened 3 years ago

zhDai commented 3 years ago

Hello, I have a question about the functions of predict and predict_proba. When I run the codes like: fine_labels = clf.predict(fine_cells) fine_probs = clf.predict_proba(fine_cells) print(fine_labels ) print(fine_probs) the outcomes is: 0.0 [0.00393306 0.88768286 0.00303451 0.00343445 0.00206198 0.07225432 0.00329827 0.00387538 0.0039457 0.00248316 0.00339068 0.00342552 0.00270422 0.00284276 0.00163315] I think the 0.0 is the label and it is ok. However, if I choose the max Probability, the idx is 1,but it is wrong. So how can I get the accurate Probability