本文轉自：http://www.shareditor.com/blogshow/?blogId=60

任何一種機器學習模型都附帶很多引數，不同場景對應不同的最佳引數，手工嘗試各種引數無疑浪費很多時間，scikit-learn幫我們實現了自動化，那就是網格搜尋 。

網格搜尋

用法舉例

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model.logistic import LogisticRegression
from sklearn.grid_search import GridSearchCV
from sklearn.pipeline import Pipeline

# 構造樣本，這塊得多構造點，不然會報class不足的錯誤，因為gridsearch會拆分成小組
X = []
X.append("fuck you")
X.append("fuck you all")
X.append("hello everyone")
X.append("fuck me")
X.append("hello boy")
X.append("fuck you")
X.append("fuck you all")
X.append("hello everyone")
X.append("fuck me")
X.append("hello boy")
X.append("fuck you")
X.append("fuck you all")
X.append("hello everyone")
X.append("fuck me")
X.append("hello boy")
X.append("fuck you")
X.append("fuck you all")
X.append("hello everyone")
X.append("fuck me")
X.append("hello boy")
X.append("fuck you")
X.append("fuck you all")
X.append("hello everyone")
X.append("fuck me")
X.append("hello boy")

y = [1,0,1,0,1,1,0,1,0,1,1,0,1,0,1,1,0,1,0,1,1,0,1,0,1]

# 這是執行的序列，gridsearch是構造多程序順序執行序列並比較結果
# 這裡的vect和clf名字自己隨便起，但是要和parameters中的字首對應
pipeline = Pipeline([
    ('vect', TfidfVectorizer(stop_words='english')),
    ('clf', LogisticRegression())
    ])

# 這裡面的max_features必須是TfidfVectorizer的引數, 裡面的取值就是子程序分別執行所用
parameters = {
        'vect__max_features': (3, 5),
        }

# accuracy表示按精確度判斷最優值
grid_search = GridSearchCV(pipeline, parameters, n_jobs = -1, verbose = 1, scoring = 'accuracy', cv = 3)
grid_search.fit(X, y)

print '最佳效果: %0.3f' % grid_search.best_score_
print '最優引數組合: '
best_parameters = grid_search.best_estimator_.get_params()
for param_name in sorted(parameters.keys()):
    print('\t%s: %r' % (param_name, best_parameters[param_name]))