3

import pandas as pd
train_df = pd.read_csv('1/train_set.csv', sep='\t', nrows=15000)

train_df.head()

	label	text
0	2	2967 6758 339 2021 1854 3731 4109 3792 4149 15...
1	11	4464 486 6352 5619 2465 4802 1452 3137 5778 54...
2	3	7346 4068 5074 3747 5681 6093 1777 2226 7354 6...
3	2	7159 948 4866 2109 5520 2490 211 3956 5520 549...
4	3	3646 3055 3055 2490 4659 6065 3370 5814 2465 5...

Count Vectors + RidgeClassifier

import pandas as pd

from sklearn.feature_extraction.text import CountVectorizer
from sklearn.linear_model import RidgeClassifier
from sklearn.metrics import f1_score

vectorizer = CountVectorizer(max_features=3000)
train_test = vectorizer.fit_transform(train_df['text'])

clf = RidgeClassifier()
clf.fit(train_test[:10000], train_df['label'].values[:10000])

val_pred = clf.predict(train_test[10000:])
print(f1_score(train_df['label'].values[10000:], val_pred, average='macro'))
# 0.74

0.65441877581244

TF-IDF + RidgeClassifier

from sklearn.feature_extraction.text import TfidfVectorizer

tfidf = TfidfVectorizer(ngram_range=(1,3), max_features=3000)
train_test = tfidf.fit_transform(train_df['text'])

clf = RidgeClassifier()
clf.fit(train_test[:10000], train_df['label'].values[:10000])

val_pred = clf.predict(train_test[10000:])
print(f1_score(train_df['label'].values[10000:], val_pred, average='macro'))
# 0.87

0.8719098297954606

尝试改变TF-IDF的参数，并验证精度

f1 ={}

def idf(x):
    tfidf = TfidfVectorizer(ngram_range=x, max_features=3000)
    train_test = tfidf.fit_transform(train_df['text'])

    clf = RidgeClassifier()
    clf.fit(train_test[:10000], train_df['label'].values[:10000])

    val_pred = clf.predict(train_test[10000:])
    f = f1_score(train_df['label'].values[10000:], val_pred, average='macro')
    f1[x] = f

for a in range(1,5):
    for b in range(1,6):
        if a <=b:
            x = (a,b)
            idf(x)

f1

{(1, 1): 0.8591560167309494,
 (1, 2): 0.8720847996197074,
 (1, 3): 0.8719372173702,
 (1, 4): 0.8736955495856327,
 (1, 5): 0.8753973492633325,
 (2, 2): 0.8431077261676884,
 (2, 3): 0.8247590162358751,
 (2, 4): 0.8217618443321154,
 (2, 5): 0.8254233297052576,
 (3, 3): 0.7544838165811477,
 (3, 4): 0.7579950226623512,
 (3, 5): 0.7388010943827312,
 (4, 4): 0.6651324921251808,
 (4, 5): 0.6346451598446852}

f1 = {}
for b in range(1,20):
    x = (1,b)
    idf(x)
f1

尝试使用其他机器学习模型，完成训练和验证

from sklearn import svm
clf=svm.SVC()

tfidf = TfidfVectorizer(ngram_range=(1,5), max_features=3000)
train_test = tfidf.fit_transform(train_df['text'])
clf.fit(train_test[:10000], train_df['label'].values[:10000])
val_pred = clf.predict(train_test[10000:])
print(f1_score(train_df['label'].values[10000:], val_pred, average='macro'))

import sklearn.model_selection as sk_model_selection
accs=sk_model_selection.cross_val_score(model, iris_X, y=iris_y, scoring=None,cv=10, n_jobs=1)
print('交叉验证结果:',accs)