Scikit-learn算法总结

Scikit-learn算法总结

1 线性模型

  主要在sklearn.linear_model包下

普通线性回归模型：

• LinearRegression ： f(x) = w · x + b

  原型：class sklearn.linear_model.LinearRegression(fit_intercept=True, normalize=False， copy_X=True, n_jobs=1)

from sklearn.linear_model import LinearRegression
from sklearn import datasets, model_selection

# 导入数据回归，数据为sklearn自带的糖尿病人采集数据
def load_data_regression():
    '''
    :function:导入数据：数据为sklearn自带的糖尿病人采集数据
    :return: model_selection.train_test_split函数可以将数据集分为训练集和测试集；
             返回一个元组    X_train, X_test, y_train, y_test
    '''
    diabetes = datasets.load_diabetes()
    diabetes_X = diabetes.data
    diabetes_y = diabetes.target
    print(diabetes_X)
    print(diabetes_y)
    return model_selection.train_test_split(diabetes_X, diabetes_y, test_size=0.25, random_state=0)

#普通线性回归预测
def test_LinearRegression(X_train, X_test, y_train, y_test):
    regr = LinearRegression()
    regr.fit(X_train, y_train)
    print('Score:%.2f' %regr.score(X_test, y_test))

X_train, X_test, y_train, y_test = load_data_regression()
test_LinearRegression(X_train, X_test, y_train, y_test)

普通线性回归的预测值很低 Score:0.36

逻辑回归模型：

• LogisticRegression ： f(x) = w · x + b

  原型：class sklearn.linear_model.LogisticRegression(penalty=’l2’, dual=False, tol=0.0001, C=1.0, fit_intercept=True, intercept_scaling=1,class_weight=None, solver=’liblinear’, max_iter=100, multi_class=’ovr’, verbose=0, n_jobs=1)

  multi_class：一个字符串，指定对于多分类问题的策略，

   - ‘ovr’：采用one-vs-rest策略
   - ‘multinomial’：直接采用多分类逻辑回归策略
  

  C：C是正则化项系数的倒数，它越小则正则化项的权重越大

from sklearn.linear_model import LogisticRegression
from sklearn import datasets, model_selection

# 导入分类数据，数据为sklearn自带的手写识别数据集digit dataset
def load_data_classification():
    '''
    :function:导入数据：数据为sklearn自带的手写识别数据集digit dataset
    :return: model_selection.train_test_split函数可以将数据集分为训练集和测试集；
             返回一个元组    X_train, X_test, y_train, y_test
    '''
    digits = datasets.load_digits()
    digits_X = digits.data
    digits_y = digits.target
    return model_selection.train_test_split(digits_X, digits_y, test_size=0.25, random_state=0)

#逻辑回归
def test_LogisticRegression(X_train, X_test, y_train, y_test):
    regr1 = LogisticRegression(multi_class='multinomial', solver='lbfgs')
    regr1.fit(X_train, y_train)
    print('Score:%.2f' %regr1.score(X_test, y_test))

X_train, X_test, y_train, y_test = load_data_c