CODEWITHSUNDEEP
pythonpython-3.xscikit-learnclassification
Sidra Idrees
Sidra Idrees

Reputation: 41

sklearn SGDClassifier, results a label when no match available?

I am using sklearn SGDClassifier for predicting my data set. I have text which should not be classified in any label/category, but I am amazed, if I am giving test data as "kjhd askdhajksdh asd askh", it is still getting classified to one of the given categories.

I have worked with probabilities as well but still having a valuable probability for this junk text.

My question is, can classifier return something like "No match found" etc for such cases?

Upvotes: 0

Views: 173

Answers (1)

Vivek Kumar
Vivek Kumar

Reputation: 36609

No. Classifier will classify any input to one of the labels in training with the highest probability, however small are these probabilities.

You can use decision_function for setting a threshold of a label. Something like: 

threshold = 0.25
if confidence_score < threshold:
    print("No match found")

Upvotes: 1

Related Questions