sklearn 机器学习库案例

左手的ㄟ右手 2022-03-26 15:30 317阅读 0赞

sklearn是非常流行的机器学习库，实现了很多的机器学习模型。官网： [http://scikit-learn.org/stable/][http_scikit-learn.org_stable]  里面有全面的实例和模型参数讲解，用到哪个模型就去官方查看说明文档。

基本功能主要被分为六大部分：分类，回归，聚类，数据降维，模型选择和数据预处理。

Estimator框架的基本使用套路：

model = EstimatorObject()  \#得到模型

model.fit(dataset.data, dataset.target)   \#训练模型

model.predict(dataser.data)    \#预测

本文对主要的机器学习模型进行实例演示，具体模型的参数结合的自己需求设置。

1.分类问题

数据集为 Car Ecaluation,根据汽车的若干属性对汽车性能进行评价。下载地址： [http://archive.ics.uci.edu/ml/datasets/Car+Evaluation][http_archive.ics.uci.edu_ml_datasets_Car_Evaluation]

预处理：将数据集保存后将后缀直接改为csv，并将里面用字符串表示的等级转化为数字。如small,low,unacc转化为1，2，3

1.1 SVM支持向量机模型

from sklearn import svm 
         import pandas as pd 
         import numpy as np 
         from sklearn.model_selection import train_test_split 
         #SVM模型实现汽车性能评测 
         car_data = pd.read_csv( r'D:\pyproject\sklearn\car.csv') 
         car_data = car_data.dropna() #去掉缺失值 
         #提取特征和类别 
         X= car_data.ix[:, : 'safety'] 
         y= car_data.ix[:, 'class'] 
         #划分训练集和测试集 
         X_train, X_test, y_train, y_test = train_test_split(X, y, test_size= 0.3, random_state= 0) 
         # 建立模型。 设置算法内核类型，有 'linear’, ‘poly’, ‘rbf’, ‘sigmoid’;惩罚参数为1，一般为10的幂次方 
         svc_model = svm.SVC(kernel= 'rbf', C= 1) 
         svc_model.fit(X_train, y_train) 
         predict_data = svc_model.predict(X_test) 
         accuracy = np.mean(predict_data==y_test) 
         print(accuracy)

运行结果：

!\[在这里插入图片描述\](https://img-blog.csdn.net/20180502175729629?watermark/2/text/aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzI3MTUwODkz/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70)

1.2 MLP神经网络模型

from sklearn.neural_network import MLPClassifier 
         import pandas as pd 
         import numpy as np 
         from sklearn.model_selection import train_test_split 
         #MLP神经网络模型实现汽车性能评测 
         car_data = pd.read_csv( r'D:\pyproject\sklearn\car.csv') 
         car_data = car_data.dropna() #去掉缺失值 
         #提取特征和对象类别 
         X= car_data.ix[:, : 'safety'] 
         y= car_data.ix[:, 'class'] 
         #划分训练集和测试集 
         X_train, X_test, y_train, y_test = train_test_split(X, y, test_size= 0.3, random_state= 0) 
         #建立MLP神经网络模型 ，MLP的求解方法为adam，可选lbfgs、sgd，正则化惩罚alpha = 0.1 
         mpl_model = MLPClassifier(solver= 'adam', learning_rate= 'constant', learning_rate_init= 0.01,max_iter = 500,alpha = 0.01) 
         mpl_model.fit(X_train, y_train) 
         predict_data = mpl_model.predict(X_test) 
         accuracy = np.mean(predict_data == y_test) 
         print(accuracy)

运行结果：

!\[在这里插入图片描述\](https://img-blog.csdn.net/20180502175832180?watermark/2/text/aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzI3MTUwODkz/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70)

1.3 逻辑回归模型

import pandas as pd 
         import numpy as np 
         from sklearn.model_selection import train_test_split 
         from sklearn.linear_model import LogisticRegression 
         #逻辑回归模型实现汽车性能预测 
         car_data = pd.read_csv( r'D:\pyproject\sklearn\car.csv') 
         car_data = car_data.dropna() #去掉缺失值 
         #提取特征和对象类别 
         X= car_data.ix[:, : 'safety'] 
         y= car_data.ix[:, 'class'] 
         #划分训练集和测试集 
         X_train, X_test, y_train, y_test = train_test_split(X, y, test_size= 0.3, random_state= 0) 
         #建立逻辑回归模型 ，惩罚参数为100 
         lr_model = LogisticRegression(C= 100, max_iter= 1000) 
         lr_model.fit(X_train, y_train) 
         predict_data = lr_model.predict(X_test) 
         accuracy = np.mean(predict_data == y_test) 
         print(accuracy)

运行结果：

!\[在这里插入图片描述\](https://img-blog.csdn.net/20180502180001759?watermark/2/text/aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzI3MTUwODkz/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70)

1.4 决策树模型

from sklearn import tree 
         import pandas as pd 
         import numpy as np 
         from sklearn.model_selection import train_test_split 
         #决策树模型实现汽车性能预测 
         car_data = pd.read_csv( r'D:\pyproject\sklearn\car.csv') 
         car_data = car_data.dropna() #去掉缺失值 
         #提取特征和类别 
         X= car_data.ix[:, : 'safety'] 
         y= car_data.ix[:, 'class'] 
         #划分训练集和测试集 
         X_train, X_test, y_train, y_test = train_test_split(X, y, test_size= 0.3, random_state= 0) 
         # 建立决策树模型，选择算法为熵增益，可选gini,entropy,默认为gini 
         tree_model = tree.DecisionTreeClassifier(criterion= 'gini') 
         tree_model.fit(X_train, y_train) 
         predict_data = tree_model.predict(X_test) 
         accuracy = np.mean(predict_data==y_test) 
         print(accuracy)

运行结果：

!\[在这里插入图片描述\](https://img-blog.csdn.net/20180502180111950?watermark/2/text/aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzI3MTUwODkz/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70)

1.5 KNN（K最临近模型）

from sklearn import neighbors 
         import pandas as pd 
         import numpy as np 
         from sklearn.model_selection import train_test_split 
         #K最邻模型实现汽车性能预测 
         car_data = pd.read_csv( r'D:\pyproject\sklearn\car.csv') 
         car_data = car_data.dropna() #去掉缺失值 
         #提取特征和类别 
         X= car_data.ix[:, : 'safety'] 
         y= car_data.ix[:, 'class'] 
         #划分训练集和测试集 
         X_train, X_test, y_train, y_test = train_test_split(X, y, test_size= 0.3, random_state= 0) 
         # 建立KNN模型，邻居数选为7，默认为5 
         knn_model = neighbors.KNeighborsClassifier(n_neighbors = 7) 
         knn_model.fit(X_train, y_train) 
         #对测试集进行预测 
         predict_data = knn_model.predict(X_test) 
         accuracy = np.mean(predict_data==y_test) 
         print(accuracy)

运行结果：

!\[在这里插入图片描述\](https://img-blog.csdn.net/20180502180212995?watermark/2/text/aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzI3MTUwODkz/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70)

2. 回归问题

这里使用sklearn自带的数据集，数据集为波斯顿房价，根据波斯顿地区若干指标对房价进行预测。

2.1 线性回归模型实现

from sklearn.linear_model import LinearRegression 
         from sklearn.datasets import load_boston 
         from sklearn.model_selection import train_test_split 
         #导入结果评价包 
         from sklearn.metrics import mean_absolute_error 
         #利用线性回归模型预测波斯顿房价 
         
         #下载sklearn自带的数据集 
         data = load_boston() 
         #建立线性回归模型 
         clf = LinearRegression() 
         #划分训练集和测试集 
         X_train, X_test, y_train, y_test = train_test_split(data.data, data.target, test_size= 0.3, random_state= 0) 
         clf.fit(X_train, y_train) 
         predict_data = clf.predict(X_test) 
         print(predict_data) 
         #平均绝对值误差对结果进行评价 
         appraise = mean_absolute_error(y_test, predict_data) 
         print(appraise)

运行结果：

!\[在这里插入图片描述\](https://img-blog.csdn.net/20180502180307872?watermark/2/text/aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzI3MTUwODkz/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70)

[http_scikit-learn.org_stable]: http://scikit-learn.org/stable/
[http_archive.ics.uci.edu_ml_datasets_Car_Evaluation]: http://archive.ics.uci.edu/ml/datasets/Car+Evaluation