CODEWITHSUNDEEP
pythonmachine-learningscikit-learnsvmtext-classification
suganthan sivananthan
suganthan sivananthan

Reputation: 143

Predict multi class in svm

I have user review dataset like

review-1, 0,1,1,0,0

review-1 is user review and 0,1,1,0,0 is review categories. one review can have multiple categories. I want to predict categories to reviews. so I implement the code that

transformer = TfidfVectorizer(lowercase=True, stop_words=stop, max_features=500)
X = transformer.fit_transform(df.Review)

X_train, X_test, y_train, y_test = train_test_split(X, df.iloc[:, 1:6],
                                                test_size=0.25, random_state=42)

SVM = svm.SVC()
SVM.fit(X_train, y_train)

But I'm getting error like

ValueError: bad input shape (75, 5)

Could anyone suggest any good solution to solve this?

Upvotes: 1

Views: 652

Answers (1)

Venkatachalam
Venkatachalam

Reputation: 16966

You could use a binary classifier (like svm.SVC()) to solve the multi-label classification problem using OneVsRestClassifier.

Example:

from sklearn.multiclass import OneVsRestClassifier

from sklearn.svm import SVC

cls = OneVsRestClassifier(estimator=SVC(gamma ='auto'))

import numpy as np
cls.fit(np.random.rand(20,10),np.random.binomial(1,0.2,size=(20,5)))

Upvotes: 5

Related Questions