Скачать или смотреть Machine Learning: Naive Bayes text classifier model

6Benches.in the best among all machine learning training institutes in Bangalore for freshers and experienced, offers 3 months machine learning classes in bangalore.

This video covers Naive Bayes classification algorithm
------------------------------------------------------------------
In the previous video, we have covered, text vectorization, that is converting unstructured text data to structured numerical representation which can be later fed to ML classification algorithm.

We shall try to fit the best model for the vectorized data for 20newsgroups dataset.

1)Couple of algorithms that we can use are
K nearest neighbors classifier, svm, naive bayes classifier.
Naive bayes classifier is covered in this class.

2) Three types of naive bayes algorithm,
a)Guassian: used when data is normally distributed over particular mean value
b) Multinomial: used when countvectorizer used instead of tfidf vectorizer for discrete counts/frequency of count
c)Bernoulli used when feature vectors are binary(one hot encoding)

3) from sklearn.naive_bayes import MultinomialNB, BernoulliNB, GaussianNB
3a) text vectorization
from sklearn.feature_extraction.text import TfidfVectorizer
vectorizer = TfidfVectorizer()
x = vectorizer.fit_transform(text)
from sklearn.datasets import fetch_20newsgroups
twenty_train = fetch_20newsgroups(subset = 'train', shuffle = true)
y=twenty_train.target

3b) fit naive bayes model
clf = MultinomialNB()
y_prediction = clf.fit(x[0:5000],y[0:5000]) #training on 5000 records

y_actual = clf.predict(x[5000:5100]) #predicting for 100 records

3c) calculate accuracy of naive bayes model
from sklearn.metrics import accuracy_score
accuracy_score(y_actual, y_prediction)