8. advance.py

from reviews import counter, training_counts
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.naive_bayes import MultinomialNB

# Add your review:
review = "its good"
review_counts = counter.transform([review])

classifier = MultinomialNB()
training_labels = [0] * 1000 + [1] * 1000

classifier.fit(training_counts, training_labels)
  
neg = (classifier.predict_proba(review_counts)[0][0] * 100).round()
pos = (classifier.predict_proba(review_counts)[0][1] * 100).round()

if pos > 50:
  print("Thank you for your positive review!")
elif neg > 50:
  print("We're sorry this hasn't been the best possible lesson for you! We're always looking to improve.")
else:
  print("Naive Bayes cannot determine if this is negative or positive. Thank you or we're sorry?")

  
print("\nAccording to our trained Naive Bayes classifier, the probability that your review was negative was {0}% and the probability it was positive was {1}%.".format(neg, pos))