import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
from sklearn.preprocessing import LabelEncoder
from sklearn.neighbors import KNeighborsClassifier
from sklearn.decomposition import PCA
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
#加载数据返回dataframe
def load_data():
data = pd.read_csv('D:\data\lgdata.csv',index_col=False, header=None,names=['target','x1','x2','x3','x4','x5',
'x6','x7','x8','x9','x10','x11','x12',
'x13','x14','x15','x16','x17','x18','x19',
'x20','x21','x22'])
return data
#特征工程：补齐缺失值对离散特征进行独热码以及对target进行labelecode
def deal_feature(data):
#首先使用knn对特征进行补齐
#在此之前先将数据集进行独热码处理，否则无法使用knn
#需要补齐得数据
l_data = data.loc[data['x11'].isin(['?'])]
#训练数据
t_data = data.loc[data['x11']!='?']
#训练数据得x
t_x = t_data.loc[:,['target','x1','x2','x3','x4','x5',
'x6','x7','x8','x9','x10','x12',
'x13','x14','x15','x16','x17','x18','x19',
'x20','x21','x22']]
#独热码,实际是增加了维度
t_x = pd.get_dummies(t_x)
#训练得y
t_y = t_data.loc[:,['x11']]
#对目标labelencoder处理
x11_l = LabelEncoder()
#处理训练集得y
t_y = x11_l.fit_transform(t_y)
#需要补齐得数据，测试集
l_x = l_data.loc[:,['target','x1','x2','x3','x4','x5',
'x6','x7','x8','x9','x10','x12',
'x13','x14','x15','x16','x17','x18','x19',
'x20','x21','x22']]
l_data = l_x
l_x = pd.get_dummies(l_x)
#由于数据不随机性导致。特征值缺失，所以只能将训练集和测试集得共同特征提取出来
f = []
#之对共同特征进行处理
f = l_x.columns&t_x.columns
t_x = t_x.loc[:,f]
#首先将特征转化为独热码以及x11转化为lable
knn = KNeighborsClassifier(n_neighbors=5)
knn.fit(t_x,t_y)
l_x = l_x.loc[:,f]
x11 = knn.predict(l_x)
x11 = x11_l.inverse_transform(x11)
#将x11放到data中
# 形成新得dataframe
# t_data和l_data,x11,
l_data.insert(11, 'x11', x11)
#data为填充好得数据
data = t_data.append(l_data)
# 首先对target进行类别编码
target_le = LabelEncoder()
y = target_le.fit_transform(data['target'].values)
a = ['x1', 'x2', 'x3', 'x4', 'x5',
'x6', 'x7', 'x8', 'x9', 'x10', 'x11', 'x12',
'x13', 'x14', 'x15', 'x16', 'x17', 'x18', 'x19',
'x20', 'x21', 'x22']
# 对离散化特征独热码处理
x = pd.get_dummies(data[a])
return x,y
def deal_pca(data):
pca = PCA(n_components=70)
t_data = pca.fit_transform(data)
print(pca.explained_variance_)
return t_data
def test_L(x,y):
x_train, x_test, y_train, y_test = train_test_split(x, y, train_size=0.7, random_state=2)
print(x_train.shape)
print(x_test.shape)
# 逻辑回归。写出损失函数
LR = LogisticRegression(C=1.0, penalty='l1', tol=0.01)
LR.fit(x_train, y_train)
LR.predict(x_test)
print(LR.score(x_test, y_test))
if __name__=='__main__':
#处理特征
x,y = deal_feature(load_data())
test_L(x,y)
#pca降维处理
x = deal_pca(x)
test_L(x,y)
#画图
'''
data = pd.DataFrame(x,columns=['x1','x2','x3'])
print(data)
data.insert(3,'class',y)
print(data)
positive = data[data['class'] == 1]
negative = data[data['class'] == 0]
fig = plt.figure(figsize=(100,100))
ax = Axes3D(fig)
ax.scatter(positive['x1'], positive['x2'],positive['x3'], c='r', marker='x', s=30, label='p')
ax.scatter(negative['x1'], negative['x2'], negative['x3'],s=30, c='b', marker='o', label='e')
ax.legend()
ax.set_xlabel('x1')
ax.set_ylabel('x2')
ax.set_zlabel('x3')
plt.show()
'''
#划分训练集,测试集
x_train, x_test, y_train, y_test = train_test_split(x, y, train_size=0.7, random_state=1)
print(x_train.shape)
print(x_test.shape)
#逻辑回归。写出损失函数
#梯度下降法求训练参数
