감성 분석 (3)

8.3 학습을 통한 머신러닝 기반의 감성 분석

8.3.1 NLTK 영화 리뷰에 대한 머신러닝 기반 감성 분석

from sklearn.model_selection import train_test_split #sklearn에서 제공하는 split 함수를 사용

X_train, X_test, y_train, y_test = train_test_split(reviews, categories, test_size=0.2, random_state=7)

print('Train set count: ', len(X_train))
print('Test set count: ', len(X_test))

"""
Train set count:  1600
Test set count:  400
/usr/local/lib/python3.10/dist-packages/ipykernel/ipkernel.py:283: DeprecationWarning: `should_run_async` will not call `transform_cell` automatically in the future. Please pass the result to `transformed_cell` argument and any exception that happen during thetransform in `preprocessing_exc_tuple` in IPython 7.17 and above.
  and should_run_async(code)
"""

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.naive_bayes import MultinomialNB #sklearn이 제공하는 MultinomialNB 를 사용

tfidf = TfidfVectorizer().fit(X_train) 

X_train_tfidf = tfidf.transform(X_train) # train set을 변환
print('#Train set dimension:', X_train_tfidf.shape) # 실제로 몇개의 특성이 사용되었는지 확인
X_test_tfidf = tfidf.transform(X_test) # test set을 변환
print('#Test set dimension:', X_test_tfidf.shape)

NB_clf = MultinomialNB(alpha=0.01) # 분류기 선언
NB_clf.fit(X_train_tfidf, y_train) #train set을 이용하여 분류기(classifier)를 학습
print('#Train set score: {:.3f}'.format(NB_clf.score(X_train_tfidf, y_train))) #train set에 대한 예측정확도를 확인
print('#Test set score: {:.3f}'.format(NB_clf.score(X_test_tfidf, y_test))) #test set에 대한 예측정확도를 확인

"""
/usr/local/lib/python3.10/dist-packages/ipykernel/ipkernel.py:283: DeprecationWarning: `should_run_async` will not call `transform_cell` automatically in the future. Please pass the result to `transformed_cell` argument and any exception that happen during thetransform in `preprocessing_exc_tuple` in IPython 7.17 and above.
  and should_run_async(code)
#Train set dimension: (1600, 36189)
#Test set dimension: (400, 36189)
#Train set score: 0.998
#Test set score: 0.797
"""

8.3.2 다음 영화 리뷰에 대한 머신러닝 기반 감성 분석

import pandas as pd
df = pd.read_csv('./data/daum_movie_review.csv')
df.head(5)

import matplotlib.pyplot as plt
%matplotlib inline

df.rating.value_counts().sort_index().plot(kind='bar')
#df.rating.plot.hist(bins=10) #히스토그램을 그릴 수도 있다.
plt.show()

• 클래스별로 고르게 분포하지 않은 경우를 불균형 데이터셋이라 함
• 이 때는 성능이 잘 나오기 어렵고 모형의 성능을 정확도로 측정하기 어려움
• 불균형 데이터의 성능 측정을 위해 정밀도와 재현율 사용
• 정밀도: 긍정으로 예측한 대상 중 실제로 긍정인 데이터 비율
• 재현율: 실제 긍적인 데이터 중 긍정으로 예측한 대상의 비율

from sklearn.model_selection import train_test_split

# split data and labels into a training and a test set
X_train, X_test, y_train, y_test = train_test_split(df.review, df.rating, random_state=7)
print('#Train set size:', len(X_train))
print('#Test set size:', len(X_test))

from konlpy.tag import Okt #konlpy에서 Twitter 형태소 분석기를 import
#from konlpy.tag import Twitter #konlpy에서 Twitter 형태소 분석기를 import
okt = Okt()

def twit_tokenizer(text): #전체를 다 사용하는 대신, 명사, 동사, 형용사를 사용
    target_tags = ['Noun', 'Verb', 'Adjective']
    result = []
    for word, tag in okt.pos(text, norm=True, stem=True):
        if tag in target_tags:
            result.append(word)
    return result

tfidf = TfidfVectorizer(tokenizer=twit_tokenizer, max_features=2000, min_df=5, max_df=0.5) #명사, 동사, 형용사를 이용하여 tfidf 생성
X_train_tfidf = tfidf.fit_transform(X_train)
X_test_tfidf = tfidf.transform(X_test)

"""
/usr/local/lib/python3.10/dist-packages/ipykernel/ipkernel.py:283: DeprecationWarning: `should_run_async` will not call `transform_cell` automatically in the future. Please pass the result to `transformed_cell` argument and any exception that happen during thetransform in `preprocessing_exc_tuple` in IPython 7.17 and above.
  and should_run_async(code)
/usr/local/lib/python3.10/dist-packages/sklearn/feature_extraction/text.py:528: UserWarning: The parameter 'token_pattern' will not be used since 'tokenizer' is not None'
  warnings.warn(
#Train set size: 11043
#Test set size: 3682
"""

from sklearn.linear_model import LinearRegression

lr = LinearRegression()  #객체를 생성
lr.fit(X_train_tfidf, y_train)
print('#Regression Train set R2 score: {:.3f}'.format(lr.score(X_train_tfidf, y_train)))
print('#Regression Test set R2 score: {:.3f}'.format(lr.score(X_test_tfidf, y_test)))

"""
/usr/local/lib/python3.10/dist-packages/ipykernel/ipkernel.py:283: DeprecationWarning: `should_run_async` will not call `transform_cell` automatically in the future. Please pass the result to `transformed_cell` argument and any exception that happen during thetransform in `preprocessing_exc_tuple` in IPython 7.17 and above.
  and should_run_async(code)
#Regression Train set R2 score: 0.605
#Regression Test set R2 score: 0.395
"""

y_train_senti = (y_train > 5)
y_test_senti = (y_test > 5)

y_train_predict = (lr.predict(X_train_tfidf) > 5)
y_test_predict = (lr.predict(X_test_tfidf) > 5)

from sklearn.metrics import precision_score
from sklearn.metrics import recall_score
from sklearn.metrics import f1_score

print('#Accuracy for train set: {:.3f}'.format(accuracy_score(y_train_senti, y_train_predict)))
print('#Precision for train set: {:.3f}'.format(precision_score(y_train_senti, y_train_predict)))
print('#Recall for train set: {:.3f}'.format(recall_score(y_train_senti, y_train_predict)))
print('#F1 for train set: {:.3f}'.format(f1_score(y_train_senti, y_train_predict)))

print('#Accuracy for test set: {:.3f}'.format(accuracy_score(y_test_senti, y_test_predict)))
print('#Precision for test set: {:.3f}'.format(precision_score(y_test_senti, y_test_predict)))
print('#Recall for test set: {:.3f}'.format(recall_score(y_test_senti, y_test_predict)))
print('#F1 for test set: {:.3f}'.format(f1_score(y_test_senti, y_test_predict)))

"""
#Accuracy for train set: 0.888
#Precision for train set: 0.893
#Recall for train set: 0.969
#F1 for train set: 0.929
#Accuracy for test set: 0.848
#Precision for test set: 0.868
#Recall for test set: 0.946
#F1 for test set: 0.905
"""

from sklearn.linear_model import LogisticRegression #sklearn이 제공하는 logistic regression을 사용

#count vector에 대해 regression을 해서 NB와 비교
LR_clf = LogisticRegression() #분류기 선언
LR_clf.fit(X_train_tfidf, y_train_senti) # train data를 이용하여 분류기를 학습

y_train_predict = LR_clf.predict(X_train_tfidf)
y_test_predict = LR_clf.predict(X_test_tfidf)

print('#Accuracy for train set: {:.3f}'.format(accuracy_score(y_train_senti, y_train_predict)))
print('#Precision for train set: {:.3f}'.format(precision_score(y_train_senti, y_train_predict)))
print('#Recall for train set: {:.3f}'.format(recall_score(y_train_senti, y_train_predict)))
print('#F1 for train set: {:.3f}'.format(f1_score(y_train_senti, y_train_predict)))

print('#Accuracy for test set: {:.3f}'.format(accuracy_score(y_test_senti, y_test_predict)))
print('#Precision for test set: {:.3f}'.format(precision_score(y_test_senti, y_test_predict)))
print('#Recall for test set: {:.3f}'.format(recall_score(y_test_senti, y_test_predict)))
print('#F1 for test set: {:.3f}'.format(f1_score(y_test_senti, y_test_predict)))

"""
#Accuracy for train set: 0.878
#Precision for train set: 0.878
#Recall for train set: 0.973
#F1 for train set: 0.923
#Accuracy for test set: 0.855
#Precision for test set: 0.866
#Recall for test set: 0.958
#F1 for test set: 0.910
"""

 

 

 

 

 

※ 해당 내용은 <파이썬 텍스트 마이닝 완벽 가이드>의 내용을 토대로 학습하며 정리한 내용입니다.