sklearn学习笔记

参考：https://www.bilibili.com/video/BV1xW411Y7Qd?p=8

1. 如何选择机器学习方法：sklearn官方图
   https://scikit-learn.org/stable/tutorial/machine_learning_map/index.html

2. 通用学习模式

import numpy as np 
from sklearn import datasets
from sklearn.model_selection import train_test_split 
from sklearn.neighbors import KNeighborsClassifier

## 1. 读取数据
iris = datasets.load_iris() #iris is a type of flower
iris_X = iris.data
iris_y = iris.target

# 看一下数据
#print(iris_X[:2,:]) # iris的属性 两个sample
#print(iris_y) # y的分类 输出：3个类别

## 2. 分类
X_train,X_test,y_train,y_test = train_test_split(iris_X,iris_y,test_size=0.3) # train：test = 7:3
#print(y_train) #输出可以看出分开了且打乱了数据

## 3. fit
knn = KNeighborsClassifier()
knn.fit(X_train,y_train) 

# 用预测和真实的数据对比
print(knn.predict(X_test))
print(y_test)

3. sklearn的数据库

from sklearn import datasets
from sklearn.linear_model import LinearRegression

loaded_data = datasets.load_boston()
data_X = loaded_data.data
data_y = loaded_data.target

model = LinearRegression() # 选择模型，参数此处为默认
model.fit(data_X,data_y)

print(model.predict(data_X[:4,:])) #预测前4个
print(data_y[:4]) #与前四个真实值对比

# make some datapoints
import matplotlib.pyplot as plt
X,y= datasets.make_regression(n_samples=100,n_features=1,n_targets=1,noise=1)
plt.scatter(X,y)

# noise bigger
X,y= datasets.make_regression(n_samples=100,n_features=1,n_targets=1,noise=10)
plt.scatter(X,y)

4. sklearn的属性和功能

from sklearn import datasets
from sklearn.linear_model import LinearRegression

loaded_data = datasets.load_boston()
data_X = loaded_data.data
data_y = loaded_data.target

model = LinearRegression() # 选择模型，参数此处为默认
model.fit(data_X,data_y)

# LR的参数
print(model.coef_) # 每个属性配套的参数
print(model.intercept_)

# 功能
print(model.get_params) # model本身的参数
print(model.score(data_X,data_y)) #R^2 coefficient of determination 给model打分，看预测和实际数据多吻合

5. normalization(scale)：
   • make sure features are on a similar scale
   • 如果x1,x2取值范围跨度很大，不容易走到gradient descent的中间点

from sklearn import preprocessing
import numpy as numpy

a = np.array([[10,2.7,3.6],[-100,5,-2],[120,20,40]],dtype=np.float64)
print(a)
print(preprocessing.scale(a))

from sklearn.model_selection import train_test_split 
from sklearn.datasets.samples_generator import make_classification
from sklearn.svm import SVC
import matplotlib.pyplot as plt 

# generate some data
X,y = make_classification(n_samples=300, n_features=2, n_redundant=0,n_informative=2, random_state=22, n_clusters_per_class=1,scale=100 ) # random_state随机产生的每次产生的数据是一样的

# look at data
#plt.scatter(X[:,0],X[:,1],c=y)

X = preprocessing.scale(X) #可以注释掉对比preprocessing前后的performance
X_train,X_test,y_train,y_test = train_test_split(X,y,test_size=.3)
clf = SVC()
clf.fit(X_train,y_train)
print(clf.score(X_test,y_test)) 

6. 如何检验神经网络

   • Accuracy
   • R2 score
   • f1 score

   overfitting:

   • error_training < error_test

   解决方法：

   • Theano L1/L2 regularization
   • Tensorflow dropout
   • sklearn cross validation
     怎么确定哪个参数更好解决问题？（e.g. N_layer）

7. Cross Validation

   • 对比不同参数/model/x属性进行验证

from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split 
from sklearn.neighbors import KNeighborsClassifier

iris = load_iris() 
X = iris.data
y = iris.target

X_train,X_test,y_train,y_test = train_test_split(X,y,random_state=4) 

knn = KNeighborsClassifier(n_neighbors=5) # 考虑数据点附近的5个数据
knn.fit(X_train,y_train) 
print(knn.score(X_test,y_test))

# 添加了CV
from sklearn.model_selection import cross_val_score
knn = KNeighborsClassifier(n_neighbors=5)
scores = cross_val_score(knn,X,y,cv=5,scoring = 'accuracy')
print(scores.mean())

from sklearn.model_selection import cross_val_score
import matplotlib.pyplot as plt 
k_range = range(1,31)
k_scores = []
for k in k_range:
    knn = KNeighborsClassifier(n_neighbors=k)
    #scores = cross_val_score(knn,X,y,cv=10,scoring = 'accuracy') # for classificaiton
    loss= -cross_val_score(knn,X,y,cv=10,scoring = 'neg_mean_squared_error') # for regression
    k_scores.append(loss.mean()) #换成scores

plt.plot(k_range,k_scores)
plt.xlabel('Value of K for KNN')
plt.ylabel('Cross validation accuracy')

# overfitting 问题
from sklearn.model_selection import learning_curve
from sklearn.datasets import load_digits
from sklearn.svm import SVC
import matplotlib.pyplot as plt 
import numpy as np 

digits = load_digits()
X = digits.data
y = digits.target

train_sizes, train_loss, test_loss = learning_curve(
    SVC(gamma=0.001),X,y,cv=10,scoring='neg_mean_squared_error',
    train_sizes=[0.1,0.25,0.5,0.75,1] #10%,25%点记录
)
train_loss_mean = - np.mean(train_loss,axis=1) 
test_loss_mean = -np.mean(test_loss,axis=1)
# 图像化结果
plt.plot(train_sizes,train_loss_mean,'o-',color ="r", label='training')
plt.plot(train_sizes,test_loss_mean,'o-',color='g',label='CV')
plt.xlabel("training example")
plt.ylabel("loss")
plt.legend(loc="best")

# overfitting 问题
from sklearn.model_selection import validation_curve
from sklearn.datasets import load_digits
from sklearn.svm import SVC
import matplotlib.pyplot as plt 
import numpy as np 

digits = load_digits()
X = digits.data
y = digits.target
param_range = np.logspace(-6,-2.3,5)
train_loss, test_loss = validation_curve(
    SVC(),X,y,param_name='gamma',param_range=param_range,cv=10,scoring='neg_mean_squared_error',
)
train_loss_mean = - np.mean(train_loss,axis=1) 
test_loss_mean = -np.mean(test_loss,axis=1)
# 图像化结果
plt.plot(param_range,train_loss_mean,'o-',color ="r", label='training')
plt.plot(param_range,test_loss_mean,'o-',color='g',label='CV')
plt.xlabel("gamma")
plt.ylabel("loss")
plt.legend(loc="best")

8. 读取/储存处理好的model

from sklearn import svm
from sklearn import datasets

clf = svm.SVC()
iris = datasets.load_iris()
X,y = iris.data, iris.target
clf.fit(X,y)

# method1 : pickle
import pickle
## save 
#with open('save/clf.pickle','wb') as f:
 #   pickle.dump(clf,f) # 把clf导入f
## restore
with open('save/clf.pickle','rb') as f:
    clf2 = pickle.load(f)
    print(clf2.predict(X[0:1]))

# method2 : joblib 比起pickle更快
import joblib
## save 
joblib.dump(clf,'save/clf.pkl')
## restore
clf3 = joblib.load('save/clf.pkl')
print(clf3.predict(X[0:1]))