今天,我们将更深入地学习和实现8个顶级Python机器学习算法。
让我们开始Python编程中的机器学习算法之旅。
8 Python机器学习算法 - 你必须学习
以下是Python机器学习的算法:
1。线性回归
线性回归是受监督的Python机器学习算法之一,它可以观察连续特征并预测结果。根据它是在单个变量上还是在许多特征上运行,我们可以将其称为简单线性回归或多元线性回归。
这是最受欢迎的Python ML算法之一,经常被低估。它为变量分配最佳权重以创建线ax + b来预测输出。我们经常使用线性回归来估计实际值,例如基于连续变量的房屋调用和房屋成本。回归线是拟合Y = a * X + b的最佳线,表示独立变量和因变量之间的关系。
您是否了解Python机器学习环境设置?
让我们为糖尿病数据集绘制这个图。
>>>将matplotlib.pyplot导入为plt
>>>将numpy导入为np
>>>来自sklearn导入数据集,linear_model
>>>来自sklearn.metrics import mean_squared_error,r2_score
>>>糖尿病=数据集。load_diabetes ()
>>> diabetes_X = diabetes.data [ :,np.newaxis,2 ]
>>> diabetes_X_train = diabetes_X [ : - 30 ] #splitting数据到训练和测试集
>>> diabetes_X_test = diabetes_X [ - 30 :]
>>> diabetes_y_train = diabetes.target [ : - 30 ] #splitting目标分为训练和测试集
>>> diabetes_y_test = diabetes.target [ - 30 :]
>>> regr = linear_model。LinearRegression ()#线性回归对象
>>> regr。fit (diabetes_X_train,diabetes_y_train )#Use training set训练模型
LinearRegression(copy_X = True,fit_intercept = True,n_jobs = 1,normalize = False)
>>> diabetes_y_pred = regr。预测(diabetes_X_test )#Make预测
>>> regr.coef_
阵列([941.43097333])
>>> mean_squared_error (diabetes_y_test,diabetes_y_pred )
3035.0601152912695
>>> r2_score (diabetes_y_test,diabetes_y_pred )#Variance得分
0.410920728135835
>>> plt。散射(diabetes_X_test,diabetes_y_test,color = 'lavender' )
>>> plt。情节(diabetes_X_test,diabetes_y_pred,color = 'pink' ,linewidth = 3 )
[]
>>> plt。xticks (())
([],)
>>> plt。yticks (())
([],)
>>> plt。show ()
Python机器学习算法 - 线性回归
2 Logistic回归
Logistic回归是一种受监督的分类Python机器学习算法,可用于估计离散值,如0/1,是/否和真/假。这是基于一组给定的自变量。我们使用逻辑函数来预测事件的概率,这给出了0到1之间的输出。
虽然它说'回归',但这实际上是一种分类算法。Logistic回归将数据拟合到logit函数中,也称为logit回归。让我们描绘一下。
>>>将numpy导入为np
>>>将matplotlib.pyplot导入为plt
>>>来自sklearn import linear_model
>>> XMIN,XMAX = - 7 ,7 #TEST集; 高斯噪声的直线
>>> n_samples = 77
>>> np.random。种子(0 )
>>> x = np.random。正常(size = n_samples )
>>> y = (x> 0 )。astype (np.float )
>>> x [ x> 0 ] * = 3
>>> x + =。4 * np.random。正常(size = n_samples )
>>> x = x [ :,np.newaxis ]
>>> clf = linear_model。LogisticRegression (C = 1e4 )#Classifier
>>> clf。适合(x,y )
>>> plt。图(1 ,figsize = (3 ,4 ))
<图大小与300x400 0 轴>
>>> plt。clf ()
>>> plt。散射(X。拆纱()中,Y,颜色= '薰衣草' ,ZORDER = 17 )
>>> x_test = np。linspace (- 7 ,7 ,277 )
>>> def model (x ):
返回1 / (1个+ NP。EXP (-x ))
>>> loss = model (x_test * clf.coef_ + clf.intercept_ )。拉威尔()
>>> plt。plot (x_test,loss,color = 'pink' ,linewidth = 2.5 )
[]
>>> ols = linear_model。LinearRegression ()
>>> ols。适合(x,y )
LinearRegression(copy_X = True,fit_intercept = True,n_jobs = 1,normalize = False)
>>> plt。plot (x_test,ols.coef_ * x_test + ols.intercept_,linewidth = 1 )
[]
>>> plt。axhline (。4 ,颜色= ” 0.4' )
>>> plt。ylabel ('y' )
文本(0,0.5, 'Y')
>>> plt。xlabel ('x' )
文本(0.5,0, 'X')
>>> plt。xticks (范围(- 7 ,7 ))
>>> plt。yticks ([ 0 ,0.4 ,1 ] )
>>> plt。ylim (- 。25 ,1.25 )
(-0.25,1.25)
>>> plt。XLIM (- 4 ,10 )
(-4,10)
>>> plt。图例(('Logistic回归' ,'线性回归' ),loc = '右下' ,fontsize = 'small' )
>>> plt。show ()
机器学习算法 - Logistic Regreesion
3。决策树
决策树属于受监督的Python机器学习学习,并且用于分类和回归 - 尽管主要用于分类。此模型接受一个实例,遍历树,并将重要特征与确定的条件语句进行比较。是下降到左子分支还是右分支取决于结果。通常,更重要的功能更接近根。
这种Python机器学习算法可以对分类和连续因变量起作用。在这里,我们将人口分成两个或更多个同类集。让我们看看这个算法 -
>>>来自sklearn.cross_validation import train_test_split
>>>来自sklearn.tree导入DecisionTreeClassifier
>>>来自sklearn.metrics import accuracy_score
>>>来自sklearn.metrics import classification_report
>>> def importdata ():#Importing data
balance_data = PD。read_csv ( 'https://archive.ics.uci.edu/ml/machine-learning-' +
'databases / balance-scale / balance-scale.data' ,
sep = ',' ,header = None )
print (len (balance_data ))
print (balance_data.shape )
打印(balance_data。头())
return balance_data
>>> def splitdataset (balance_data ):# Splitting 数据
x = balance_data.values [ :,1 :5 ]
y = balance_data.values [ :,0 ]
x_train,x_test,y_train,y_test = train_test_split (
x,y,test_size = 0.3 ,random_state = 100 )
返回x,y,x_train,x_test,y_train,y_test
>>> def train_using_gini (x_train,x_test,y_train ):#gining with giniIndex
clf_gini = DecisionTreeClassifier (criterion = “ gini ” ,
random_state = 100 ,max_depth = 3 ,min_samples_leaf = 5 )
clf_gini。适合(x_train,y_train )
返回clf_gini
>>> def train_using_entropy (x_train,x_test,y_train ):#Training with entropy
clf_entropy = DecisionTreeClassifier (
criterion = “entropy” ,random_state = 100 ,
max_depth = 3 ,min_samples_leaf = 5 )
clf_entropy。适合(x_train,y_train )
返回clf_entropy
>>> def 预测(x_test,clf_object ):#制作预测
y_pred = clf_object。预测(x_test )
print (f “预测值:{y_pred}” )
返回y_pred
>>> def cal_accuracy (y_test,y_pred ):#计算准确性
print (confusion_matrix (y_test,y_pred ))
打印(accuracy_score (y_test,y_pred )* 100 )
print (classification_report (y_test,y_pred ))
>>> data = importdata ()
625
(625,5)
0 1 2 3 4
0 B 1 1 1 1
1 R 1 1 1 2
2 R 1 1 1 3
3 R 1 1 1 4
4 R 1 1 1 5
>>> x,y,x_train,x_test,y_train,y_test = splitdataset (data )
>>> clf_gini = train_using_gini (x_train,x_test,y_train )
>>> clf_entropy = train_using_entropy (x_train,x_test,y_train )
>>> y_pred_gini = 预测(x_test,clf_gini )
Python机器学习算法 - 决策树
>>> cal_accuracy (y_test,y_pred_gini )
[[0 6 7]
[0 67 18]
[0 19 71]]
73.40425531914893
Python机器学习算法 - 决策树
>>> y_pred_entropy = 预测(x_test,clf_entropy )
Python机器学习算法 - 决策树
>>> cal_accuracy (y_test,y_pred_entropy )
[[0 6 7]
[0 63 22]
[0 20 70]]
70.74468085106383
Python机器学习算法 - 决策树
4。支持向量机(SVM)
SVM是一种受监督的分类Python机器学习算法,它绘制了一条划分不同类别数据的线。在这个ML算法中,我们计算向量以优化线。这是为了确保每组中最近的点彼此相距最远。虽然你几乎总会发现这是一个线性向量,但它可能不是那样的。
在这个Python机器学习教程中,我们将每个数据项绘制为n维空间中的一个点。我们有n个特征,每个特征都具有某个坐标的值。
首先,让我们绘制一个数据集。
>>>来自sklearn.datasets.samples_generator import make_blobs
>>> x,y = make_blobs (n_samples = 500 ,centers = 2 ,
random_state = 0 ,cluster_std = 0 .40 )
>>>将matplotlib.pyplot导入为plt
>>> plt。scatter (x [ :,0 ] ,x [ :,1 ] ,c = y,s = 50 ,cmap = 'plasma' )
位于0x04E1BBF0的
>>> plt。show ()
Python机器学习算法 - SVM
>>>将numpy导入为np
>>> xfit = np。linspace (- 1 ,3 0.5 )
>>> plt。scatter (X [ :,0 ] ,X [ :,1 ] ,c = Y,s = 50 ,cmap = 'plasma' )
>>>为M,B,d在[ (1 ,0.65 ,0.33 ),(0.5 ,1.6 ,0.55 ),(- 0 0.2 ,2 0.9 ,0.2 )] :
yfit = m * xfit + b
PLT。情节(xfit,yfit,' - k' )
PLT。fill_between (xfit ,yfit - d,yfit + d,edgecolor = 'none' ,
color = '#AFFEDC' ,alpha = 0.4 )
[]
[]
[]
>>> plt。XLIM (- 1 ,3.5 )
(-1,3.5)
>>> plt。show ()
Python机器学习算法 - SVM
5, 朴素贝叶斯
朴素贝叶斯是一种基于贝叶斯定理的分类方法。这假定预测变量之间的独立性。朴素贝叶斯分类器将假定类中的特征与任何其他特征无关。考虑一个水果。这是一个苹果,如果它是圆形,红色,直径2.5英寸。朴素贝叶斯分类器将说这些特征独立地促成果实成为苹果的概率。即使功能相互依赖,这也是如此。
对于非常大的数据集,很容易构建朴素贝叶斯模型。这种模型不仅非常简单,而且比许多高度复杂的分类方法表现更好。让我们建立这个。
>>>来自sklearn.naive_bayes导入GaussianNB
>>>来自sklearn.naive_bayes导入MultinomialNB
>>>来自sklearn导入数据集
>>>来自sklearn.metrics import confusion_matrix
>>>来自sklearn.model_selection import train_test_split
>>> iris =数据集。load_iris ()
>>> x = iris.data
>>> y = iris.target
>>> x_train,x_test,y_train,y_test = train_test_split (x,y,test_size = 0 .3 ,random_state = 0 )
>>> gnb = GaussianNB ()
>>> MNB = MultinomialNB ()
>>> y_pred_gnb = gnb。适合(x_train,y_train )。预测(x_test )
>>> cnf_matrix_gnb = confusion_matrix (y_test,y_pred_gnb )
>>> cnf_matrix_gnb
数组([[16,0,0],
[0,18,0],
[0,0,11]],dtype = int64)
>>> y_pred_mnb = mnb。适合(x_train,y_train )。预测(x_test )
>>> cnf_matrix_mnb = confusion_matrix (y_test,y_pred_mnb )
>>> cnf_matrix_mnb
数组([[16,0,0],
[0,0,18],
[0,0,11]],dtype = int64)
6。kNN(k-Nearest Neighbors)
这是一种用于分类和回归的Python机器学习算法 - 主要用于分类。这是一种监督学习算法,它考虑不同的质心并使用通常的欧几里德函数来比较距离。然后,它分析结果并将每个点分类到组以优化它以放置所有最接近的点。它使用其邻居k的多数票对新案件进行分类。它分配给一个类的情况是其K个最近邻居中最常见的一个。为此,它使用距离函数。
I,对整个数据集进行培训和测试
>>>来自sklearn.datasets import load_iris
>>> iris = load_iris ()
>>> x = iris.data
>>> y = iris.target
>>>来自sklearn.linear_model import LogisticRegression
>>> logreg = LogisticRegression ()
>>> logreg。适合(x,y )
LogisticRegression(C = 1.0,class_weight = None,dual = False,fit_intercept = True,
intercept_scaling = 1,max_iter = 100,multi_class ='ovr',n_jobs = 1,
penalty ='l2',random_state = None,solver ='liblinear',tol = 0.0001,
verbose = 0,warm_start = False)
>>> logreg。预测(x )
array([0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
2,1,1,1,2,1,1,1,1,1,1,1,1,1,1,1,1,1,2,2,2,1,1,
1,1,1,1,1,1,1,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2,
2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,1,2,2,
2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2]]
>>> y_pred = logreg。预测(x )
>>> len (y_pred )
150
>>>来自sklearn导入指标
>>>指标。accuracy_score (y,y_pred )
0.96
>>>来自sklearn.neighbors导入KNeighborsClassifier
>>> knn = KNeighborsClassifier (n_neighbors = 5 )
>>> knn。适合(x,y )
KNeighborsClassifier(algorithm ='auto',leaf_size = 30,metric ='minkowski',
metric_params =无,n_jobs = 1,n_neighbors = 5,p = 2,
权重=“均匀”)
>>> y_pred = knn。预测(x )
>>>指标。accuracy_score (y,y_pred )
0.9666666666666667
>>> knn = KNeighborsClassifier (n_neighbors = 1 )
>>> knn。适合(x,y )
KNeighborsClassifier(algorithm ='auto',leaf_size = 30,metric ='minkowski',
metric_params =无,n_jobs = 1,n_neighbors = 1,p = 2,
权重=“均匀”)
>>> y_pred = knn。预测(x )
>>>指标。accuracy_score (y,y_pred )
1.0
II。分裂成火车/测试
>>> x.shape
(150,4)
>>> y.shape
(150)
>>>来自sklearn.cross_validation import train_test_split
>>> x.shape
(150,4)
>>> y.shape
(150)
>>>来自sklearn.cross_validation import train_test_split
>>> x_train,x_test,y_train,y_test = train_test_split (x,y,test_size = 0.4 ,random_state = 4 )
>>> x_train.shape
(90,4)
>>> x_test.shape
(60,4)
>>> y_train.shape
(90)
>>> y_test.shape
(60)
>>> logreg = LogisticRegression ()
>>> logreg。适合(x_train,y_train )
>>> y_pred = knn。预测(x_test )
>>>指标。accuracy_score (y_test,y_pred )
0.9666666666666667
>>> knn = KNeighborsClassifier (n_neighbors = 5 )
>>> knn。适合(x_train,y_train )
KNeighborsClassifier(algorithm ='auto',leaf_size = 30,metric ='minkowski',
metric_params =无,n_jobs = 1,n_neighbors = 5,p = 2,
权重=“均匀”)
>>> y_pred = knn。预测(x_test )
>>>指标。accuracy_score (y_test,y_pred )
0.9666666666666667
>>> k_range = 范围(1 ,26 )
>>>得分= [ ]
>>> for k in k_range:
knn = KNeighborsClassifier (n_neighbors = k )
KNN。适合(x_train,y_train )
y_pred = knn。预测(x_test )
分数。追加(指标。accuracy_score (y_test,y_pred ))
>>>分数
[0.95,0.95,0.9666666666666667,0.9666666666666667,0.9666666666666667,0.9833333333333333,0.9833333333333333,0.9833333333333333,0.9833333333333333,0.9833333333333333,0.9833333333333333,0.9833333333333333,0.9833333333333333,0.9833333333333333,0.9833333333333333,0.9833333333333333,0.9833333333333333,0.9666666666666667,0.9833333333333333,0.9666666666666667,0.9666666666666667,0.9666666666666667,0.9666666666666667 0.95,0.95 ]
>>>将matplotlib.pyplot导入为plt
>>> plt。情节(k_range,分数)
[]
>>> plt。xlabel ('k代表kNN' )
文字(0.5,0,'k为kNN')
>>> plt。ylabel ('测试准确度' )
文字(0,0.5,'测试准确度')
>>> plt。show ()
Python机器学习算法 - kNN(k-Nearest Neighbors)
阅读Python统计数据 - p值,相关性,T检验,KS检验
7。K-Means
k-Means是一种无监督算法,可以解决聚类问题。它使用许多集群对数据进行分类。类中的数据点与同类组是同构的和异构的。
>>>将numpy导入为np
>>>将matplotlib.pyplot导入为plt
>>>来自matplotlib导入样式
>>>风格。使用('ggplot' )
>>>来自sklearn.cluster导入KMeans
>>> X = [ 1 ,5 ,1 0.5 ,8 ,1 ,9 ]
>>> Y = [ 2 ,8 ,1.7 ,6 ,0 0.2 ,12 ]
>>> plt。散射(x,y )
>>> x = np。阵列([ [ 1 ,2 ] ,[ 5 ,8 ] ,[ 1.5 ,1 0.8 ] ,[ 8 ,8 ] ,[ 1 ,0 0.6 ] ,[ 9 ,11 ] ] )
>>> kmeans = KMeans (n_clusters = 2 )
>>> kmeans。适合(x )
KMeans(algorithm ='auto',copy_x = True,init ='k-means ++',max_iter = 300,
n_clusters = 2,n_init = 10,n_jobs = 1,precompute_distances ='auto',
random_state =无,tol = 0.0001,verbose = 0)
>>> centroids = kmeans.cluster_centers_
>>> labels = kmeans.labels_
>>>质心
数组([[1.16666667,1.46666667],
[7.33333333,9。]])
>>>标签
数组([0,1,0,1,0,1])
>>> colors = [ 'g。' ,'r。' ,'c。' ,'呃。' ]
>>> for i in range (len (x )):
print (x [ i ] ,labels [ i ] )
PLT。plot (x [ i ] [ 0 ] ,x [ i ] [ 1 ] ,colors [ labels [ i ] ] ,markersize = 10 )
[1。2.] 0
[]
[5。8.] 1
[]
[1.5 1.8] 0
[]
[8。8.] 1
[]
[1。0.6] 0
[]
[9. 11.] 1
[]
>>> plt。scatter (centroids [ :,0 ] ,centroids [ :,1 ] ,marker = 'x' ,s = 150 ,linewidths = 5 ,zorder = 10 )
>>> plt。show ()
8。Random Forest
Random Forest是决策树的集合。为了根据其属性对每个新对象进行分类,树投票给类 - 每个树提供一个分类。投票最多的分类在Random
中获胜。
>>>将numpy导入为np
>>>将pylab导入为pl
>>> x = np.random。均匀的(1 ,100 ,1000 )
>>> y = np。log (x )+ np.random。正常(0 ,。3 ,1000 )
>>> pl。scatter (x,y,s = 1 ,label = 'log(x)with noise' )
>>> pl。情节(NP。人气指数(1 ,100 ),NP。日志(NP。人气指数(1 ,100 ))中,c = 'B' ,标记= '日志(x)的函数真' )
[]
>>> pl。xlabel ('x' )
文本(0.5,0, 'X')
>>> pl。ylabel ('f(x)= log(x)' )
文本(0,0.5, 'F(X)=日志(X)')
>>> pl。传奇(loc = 'best' )
>>> pl。标题('基本日志功能' )
文字(0.5,1,'基本日志功能')
>>> pl。show ()
Python机器学习算法 -
>>>来自sklearn.datasets import load_iris
>>>来自sklearn.ensemble导入RandomForestClassifier
>>>将pandas导入为pd
>>>将numpy导入为np
>>> iris = load_iris ()
>>> df = pd。DataFrame (iris.data,columns = iris.feature_names )
>>> df [ 'is_train' ] = np.random。均匀的(0 ,1 ,LEN (DF ))<=。75
>>> df [ 'species' ] = pd.Categorical。from_codes (iris.target,iris.target_names )
>>> df。头()
萼片长度(厘米)萼片宽度(厘米)... is_train物种
0 5.1 3.5 ...真正的setosa
1 4.9 3.0 ...真正的setosa
2 4.7 3.2 ...真正的setosa
3 4.6 3.1 ...真正的setosa
4 5.0 3.6 ...假setosa
[5行x 6列]
>>> train,test = df [ df [ 'is_train' ] == True ] ,df [ df [ 'is_train' ] == False ]
>>> features = df.columns [ :4 ]
>>> clf = RandomForestClassifier (n_jobs = 2 )
>>> y,_ = pd。factorize (train [ 'species' ] )
>>> clf。适合(火车[ 功能] ,y )
RandomForestClassifier(bootstrap = True,class_weight = None,criterion ='gini',
max_depth =无,max_features ='auto',max_leaf_nodes =无,
min_impurity_decrease = 0.0,min_impurity_split =无,
min_samples_leaf = 1,min_samples_split = 2,
min_weight_fraction_leaf = 0.0,n_estimators = 10,n_jobs = 2,
oob_score = False,random_state = None,verbose = 0,
warm_start = FALSE)
>>> preds = iris.target_names [ clf。预测(测试[ 特征] )]
>>> pd。交叉表(test [ 'species' ] ,preds,rownames = [ 'actual' ] ,colnames = [ 'preds' ] )
preds setosa versicolor virginica
实际
setosa 12 0 0
versicolor 0 17 2
virginica 0 1 15
所以,这就是Python机器学习算法教程。希望你喜欢。
因此,今天我们讨论了八个重要的Python机器学习算法。您认为哪一个最具潜力?希望大家多多关注,更多精彩的文章带给大家!
大家对大数据感兴趣的可以关注我的微信公众号:大数据技术工程师
里面每天都会分享一些精彩文章,更有大数据基础与项目实战,java面试技巧,Python学习资料等等 提供给大家免费学习,回复关键字就可以领取哦