logistic regression examples using scikit-learn

gistfile1.py

# -*- coding:utf-8 -*-
from sklearn import datasets
from sklearn import linear_model
from sklearn.metrics import f1_score
##### load data and split into train and test ####
data_digits = datasets.load_digits()
data = data_digits.data
target = data_digits.target
train_ratio = 0.8
data_num = data.shape[0]
train_num = int(data_num * train_ratio)
train_data = data[:train_num,:]
val_data = data[train_num:,:]
train_target = target[:train_num]
val_target = target[train_num:]

#### train model and test ####
#### logistic regression ####
best_res = [0, 0 ,0]
res = [0,0,0]
for i in range(1,8):
    lr_model = linear_model.LogisticRegression()
    lr_l1_model = linear_model.LogisticRegression(C=10**i,penalty='l1')
    lr_l2_model = linear_model.LogisticRegression(C=10**i,penalty='l2')
    ##### fit model####
    lr_model.fit(train_data, train_target)
    lr_l1_model.fit(train_data, train_target)
    lr_l2_model.fit(train_data, train_target)
    #### test ####
    val_predict = lr_model.predict(val_data)
    val_l1_predict = lr_l1_model.predict(val_data)
    val_l2_predict = lr_l2_model.predict(val_data)
    #### result ####
    res[0] = f1_score(val_target, val_predict) 
    res[1] = f1_score(val_target, val_l1_predict) 
    res[2] = f1_score(val_target, val_l2_predict) 
    for i in range(len(best_res)):
        if res[i] > best_res[i]:
            best_res[i] = res[i] 
    print '===================='
    print 'logistic regression, f_score : %f' % res[0]
    print 'logistic regression with l1 penalty, f_score : %f' % res[1]
    print 'logistic regression with l2 penalty, f_score: %f' % res[2]

print '===================='
print 'logistic regression, best f_score : %f' % best_res[0]
print 'logistic regression with l1 penalty, best f_score : %f' % best_res[1]
print 'logistic regression with l2 penalty, best f_score: %f' % best_res[2]