Я адаптировал следующий код из сообщения Сьюзен Ли , но допустил ошибку, когда код пытается токенизировать текст с использованием ресурсов NLTK (или, возможно, что-то не так с «векторными ключами») загружается из Интернета). Произошла ошибка в 5-м блоке кода (см. Ниже, загрузка из Интернета может занять некоторое время):
## 1. load packages and data
import logging
import pandas as pd
import numpy as np
from numpy import random
import gensim
import nltk
from nltk.corpus import stopwords
from nltk import word_tokenize
from nltk import sent_tokenize
STOPWORDS = set(stopwords.words('english'))
nltk.download('stopwords')
from sklearn.model_selection import train_test_split
from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer
from sklearn.metrics import accuracy_score, confusion_matrix
import matplotlib.pyplot as plt
import re
from bs4 import BeautifulSoup
%matplotlib inline
df = pd.read_csv('https://www.dropbox.com/s/b2w7iqi7c92uztt/stack-overflow-data.csv?dl=1')
df = df[pd.notnull(df['tags'])]
my_tags = ['java','html','asp.net','c#','ruby-on-rails','jquery','mysql','php','ios','javascript','python','c','css','android','iphone','sql','objective-c','c++','angularjs','.net']
## 2. cleaning
REPLACE_BY_SPACE_RE = re.compile('[/(){}\[\]\|@,;]')
BAD_SYMBOLS_RE = re.compile('[^0-9a-z #+_]')
STOPWORDS = set(stopwords.words('english'))
def clean_text(text):
text = BeautifulSoup(text, "lxml").text # HTML decoding
text = text.lower() # lowercase text
text = REPLACE_BY_SPACE_RE.sub(' ', text) # replace REPLACE_BY_SPACE_RE symbols by space in text
text = BAD_SYMBOLS_RE.sub('', text) # delete symbols which are in BAD_SYMBOLS_RE from text
text = ' '.join(word for word in text.split() if word not in STOPWORDS) # delete stopwors from text
return text
df['post'] = df['post'].apply(clean_text)
## 3. train test split
X = df.post
y = df.tags
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state = 42)
## 4. load keyed vectors from the web: will take a while to load
import gensim
word2vec_path = "https://s3.amazonaws.com/dl4j-distribution/GoogleNews-vectors-negative300.bin.gz"
wv = gensim.models.KeyedVectors.load_word2vec_format(word2vec_path, binary=True)
wv.init_sims(replace=True)
## 5. this is where it goes wrong
def w2v_tokenize_text(text):
tokens = []
for sent in nltk.sent_tokenize(text, language='english'):
for word in nltk.word_tokenize(sent, language='english'):
if len(word) < 2:
continue
tokens.append(word)
return tokens
train, test = train_test_split(df, test_size=0.3, random_state = 42)
test_tokenized = test.apply(lambda r: w2v_tokenize_text(r['post']), axis=1).values
train_tokenized = train.apply(lambda r: w2v_tokenize_text(r['post']), axis=1).values
X_train_word_average = word_averaging_list(wv,train_tokenized)
X_test_word_average = word_averaging_list(wv,test_tokenized)
## 6. perform logistic regression test
from sklearn.linear_model import LogisticRegression
logreg = LogisticRegression(n_jobs=1, C=1e5)
logreg = logreg.fit(X_train_word_average, train['tags'])
y_pred = logreg.predict(X_test_word_average)
print('accuracy %s' % accuracy_score(y_pred, test.tags))
print(classification_report(test.tags, y_pred,target_names=my_tags))
Обновление части 5 (за комментарии @luigigi)
## 5. download nltk and use apply() function without using lambda
import nltk
nltk.download()
from nltk.corpus import stopwords
from nltk import word_tokenize
from nltk import sent_tokenize
def w2v_tokenize_text(text):
tokens = []
for sent in nltk.sent_tokenize(text, language='english'):
for word in nltk.word_tokenize(sent, language='english'):
if len(word) < 2:
continue
tokens.append(word)
return tokens
train, test = train_test_split(df, test_size=0.3, random_state = 42)
def w2v_tokenize_text(text):
tokens = []
for sent in nltk.sent_tokenize(text, language='english'):
for word in nltk.word_tokenize(sent, language='english'):
if len(word) < 2:
continue
tokens.append(word)
return tokens
train, test = train_test_split(df, test_size=0.3, random_state = 42)
test_tokenized = test['post'].apply(w2v_tokenize_text).values
train_tokenized = train['post'].apply(w2v_tokenize_text).values
X_train_word_average = word_averaging_list(wv,train_tokenized)
X_test_word_average = word_averaging_list(wv,test_tokenized)
## now run the test
from sklearn.linear_model import LogisticRegression
logreg = LogisticRegression(n_jobs=1, C=1e5)
logreg = logreg.fit(X_train_word_average, train['tags'])
y_pred = logreg.predict(X_test_word_average)
print('accuracy %s' % accuracy_score(y_pred, test.tags))
print(classification_report(test.tags, y_pred,target_names=my_tags))
Это должно работать.