Multiple linear regression (sklearn method)

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split
from sklearn import svm

from sklearn.metrics import accuracy_score
from sklearn.preprocessing import StandardScaler


# SVR LinearSVR  Return to 
# SVC LinearSVC  classification 

#  technological process 

# 1.  get data 
data = pd.read_csv('./data.csv')

# 2.  Data exploration 
# print(data.columns)
# print(data.describe())

# 3.  Data cleaning 

#  Characteristics are divided into 3 Group 
features_mean = list(data.columns[2:12])    # Average data 
features_se = list(data.columns[12:22])     # Standard deviation data 

# ID Column delete 
data.drop('id',axis=1,inplace=True)
#  take B Benign replace with 0,M Malignant replace with 1
data['diagnosis'] = data['diagnosis'].map({
    'M':1,'B':0})

print(data.head(5))

# 4.  feature selection 
#  Purpose   Dimension reduction 
sns.countplot(data['diagnosis'],label='Count')
plt.show()

#  Heat map features_mean  Correlation between fields 
corr = data[features_mean].corr()
plt.figure(figsize=(14,14))

sns.heatmap(corr,annot=True)
plt.show()

#  feature selection   Average this group  10--→6
features_remain = ['radius_mean', 'texture_mean', 'smoothness_mean', 'compactness_mean', 'symmetry_mean','fractal_dimension_mean']

#  model training 
#  extract 30% Data as a test set 
train,test = train_test_split(data,test_size=0.3)
train_x = train[features_mean]
train_y = train['diagnosis']
test_x = test[features_mean]
test_y = test['diagnosis']

#  Data normalization 
ss = StandardScaler()
train_X = ss.fit_transform(train_x)
test_X = ss.transform(test_x)

#  establish svm classifier 
model = svm.SVC()
# Parameters 
# kernel Kernel function selection 
# 1.linear  Linear kernel function   When the data is linearly separable 
# 2.poly  Polynomial kernel function   Mapping data from low dimensional space to high dimensional space   But there are many parameters , The amount of calculation is relatively large 
# 3.rbf  Gaussian kernel   Map samples to high-dimensional space   Less parameters   Good performance   Default 
# 4.sigmoid sigmoid Kernel function   In the mapping of snake spirit network  SVM Realize multilayer neural network 

# c Penalty coefficient of objective function 
# gamma  Coefficient of kernel function   The default is the reciprocal of the sample feature number 

#  Training data 
model.fit(train_x,train_y)
# 6.  Model to evaluate 
pred = model.predict(test_x)
print(' Accuracy rate ：',accuracy_score(test_y,pred))