select * from '/datasets/built-file/built_int_export.csv'

select df_1.reviewid ,df_1.content ,case when df_1.score > 5.9 then 'positive' else (case when df_1.score < 5.0 then 'negative' else 'neutral' end) end as category from df_1

import nltk import random from nltk.corpus import stopwords from nltk.sentiment.vader import SentimentIntensityAnalyzer # set your stopwords nltk.download('stopwords') stop_words = set(stopwords.words("english")) nltk.download('punkt');

# Train Naive Bayes classifier train_set = [(document_features(d), c) for (d,c) in document] classifier = nltk.NaiveBayesClassifier.train(train_set)

# Tokenize test set df_test['content'].dropna(inplace=True) df_test['content'] = df_test['content'].astype(str) def tokenize(column): tokens = nltk.word_tokenize(column) return [w for w in tokens if w.isalpha()] df_test['tokenized'] = df_test.apply(lambda x: tokenize(x['content']),axis=1) df_test[['tokenized']].head()

# Format test set for running through classifier df_test=df_test.sample(frac=1) df_test['tuples_column'] = list(zip((df_test.tokenized), df_test.category)) document2 = df_test['tuples_column'].values.tolist() test_temp=[[y.lower() for y in sublst[0]] for sublst in document] test_temp2=[sublst[1] for sublst in document] document2=tuple(zip((test_temp), test_temp2)) test_set = [(document_features(d), c) for (d,c) in document2]

print(nltk.classify.accuracy(classifier, test_set))

classifier.show_most_informative_features(5)