数据分析入门之KNN-预测年收入

操作平台： win10, python37, jupyter
数据下载： https://www.lanzous.com/iac0omd

---

1、导入数据

import numpy as np
import pandas as pd
from sklearn.neighbors import KNeighborsClassifier
salary = pd.read_csv('../data/adults.txt')
salary.shape # 结果为(32561, 15)
salary.head() #展示前5行

	age	workclass	final_weight	education	education_num	marital_status	occupation	relationship	race	sex	capital_gain	capital_loss	hours_per_week	native_country	salary
0	39	State-gov	77516	Bachelors	13	Never-married	Adm-clerical	Not-in-family	White	Male	2174	0	40	United-States	<=50K
1	50	Self-emp-not-inc	83311	Bachelors	13	Married-civ-spouse	Exec-managerial	Husband	White	Male	0	0	13	United-States	<=50K
2	38	Private	215646	HS-grad	9	Divorced	Handlers-cleaners	Not-in-family	White	Male	0	0	40	United-States	<=50K
3	53	Private	234721	11th	7	Married-civ-spouse	Handlers-cleaners	Husband	Black	Male	0	0	40	United-States	<=50K
4	28	Private	338409	Bachelors	13	Married-civ-spouse	Prof-specialty	Wife	Black	Female	0	0	40	Cuba	<=50K

2、数据预处理

2.1、选择数据

• 选择影响薪水相关性较大的数据来作为X，进行预测薪水 y :

y = salary['salary']
X = salary.iloc[:,[0,1,3,5,6,8,9,-2,-3]]
X.head()

	age	workclass	education	marital_status	occupation	race	sex	native_country	hours_per_week
0	39	State-gov	Bachelors	Never-married	Adm-clerical	White	Male	United-States	40
1	50	Self-emp-not-inc	Bachelors	Married-civ-spouse	Exec-managerial	White	Male	United-States	13
2	38	Private	HS-grad	Divorced	Handlers-cleaners	White	Male	United-States	40
3	53	Private	11th	Married-civ-spouse	Handlers-cleaners	Black	Male	United-States	40
4	28	Private	Bachelors	Married-civ-spouse	Prof-specialty	Black	Female	Cuba	40

查看数据类型

X.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 32561 entries, 0 to 32560
Data columns (total 9 columns):
age               32561 non-null int64
workclass         32561 non-null int64
education         32561 non-null int64
marital_status    32561 non-null int64
occupation        32561 non-null int64
race              32561 non-null int64
sex               32561 non-null int64
native_country    32561 non-null int64
hours_per_week    32561 non-null int64
dtypes: int64(9)
memory usage: 2.2 MB
knn = KNeighborsClassifier()

knn.fit(X,y)

结果分析： 上面的数据大多都是字符型，不能直接进行数据运算，需要进行数据的转换！

2.2、数据转化

• 把它出现的字符数据分别用对用的值来替代

2.2.1、转化字典

workclass = X['workclass'].unique()
m = {}
for i,work in enumerate(workclass):
    m[work] = i
m

{'State-gov': 0,
 'Self-emp-not-inc': 1,
 'Private': 2,
 'Federal-gov': 3,
 'Local-gov': 4,
 '?': 5,
 'Self-emp-inc': 6,
 'Without-pay': 7,
 'Never-worked': 8}

结果分析： 用0代表职业State-gov，1代表Self-emp-not-inc，2代表Private等等。

2.2.2、数据映射

X['workclass'] = X['workclass'].map(m)
X.head()

结果分析： 现在工作机构已经被映射为对应的数字了，接下了再把其他几个也映射为对应的数字。

---

列如：

u = X['occupation'].unique()
np.argwhere(u == 'Sales')[0,0]

5

---

实例：

for col in X.columns[2:-1]:
    
    u = X[col].unique()
    
    def convert(x):
        return np.argwhere(u == x)[0,0]
    
    X[col] = X[col].map(convert)
X.head()

	age	workclass	education	marital_status	occupation	race	sex	native_country	hours_per_week
0	39	0	0	0	0	0	0	0	40
1	50	1	0	1	1	0	0	0	13
2	38	2	1	2	2	0	0	0	40
3	53	2	2	1	2	1	0	0	40
4	28	2	0	1	3	1	1	1	40

3、训练数据

3.1、切分训练集和测试集

from sklearn.model_selection import train_test_split

# X -----> y一一对应
X_train,X_test,y_train,y_test = train_test_split(X,y,test_size = 0.2)

3.2、训练并预测数据

knn = KNeighborsClassifier(n_neighbors=5) #邻近值为5，可以变化邻近值，加上weights='distance'等

knn.fit(X_train,y_train)#训练模型

y_ = knn.predict(X_test)#预测数据

result = y_ == y_test #对比测试集和预测集，返回True和False

result.mean()#求平均值，代表准确率

0.7690772301550745

4、归一化处理

4.1、最大值最小值归一化

v_min = X.min()
v_max = X.max()
X2 = (X - v_min)/(v_max - v_min)
X2.head()

	age	workclass	education	marital_status	occupation	race	sex	native_country	hours_per_week
0	0.301370	0.000	0.000000	0.000000	0.000000	0.00	0.0	0.00000	0.397959
1	0.452055	0.125	0.000000	0.166667	0.071429	0.00	0.0	0.00000	0.122449
2	0.287671	0.250	0.066667	0.333333	0.142857	0.00	0.0	0.00000	0.397959
3	0.493151	0.250	0.133333	0.166667	0.142857	0.25	0.0	0.00000	0.397959
4	0.150685	0.250	0.000000	0.166667	0.214286	0.25	1.0	0.02439	0.397959

数据预测:

# 归一化，消除属性差异
X_train,X_test,y_train,y_test = train_test_split(X2,y,test_size = 0.2)
knn = KNeighborsClassifier(n_neighbors=15,weights='distance')

knn.fit(X_train,y_train)

y_ = knn.predict(X_test)

result = y_ == y_test

result.mean()

0.8174420389989252

---

自带方法：

from sklearn.preprocessing import MinMaxScaler

m = MinMaxScaler()
X4 = m.fit_transform(X)
X4[:5]

array([[0.30136986, 0.        , 0.        , 0.        , 0.        ,
        0.        , 0.        , 0.        , 0.39795918],
       [0.45205479, 0.125     , 0.        , 0.16666667, 0.07142857,
        0.        , 0.        , 0.        , 0.12244898],
       [0.28767123, 0.25      , 0.06666667, 0.33333333, 0.14285714,
        0.        , 0.        , 0.        , 0.39795918],
       [0.49315068, 0.25      , 0.13333333, 0.16666667, 0.14285714,
        0.25      , 0.        , 0.        , 0.39795918],
       [0.15068493, 0.25      , 0.        , 0.16666667, 0.21428571,
        0.25      , 1.        , 0.02439024, 0.39795918]])

4.2、方差标准化

# Z-score
v_mean = X.mean()

v_std = X.std()

X3 = (X - v_mean)/v_std
X3.head()

	age	workclass	education	marital_status	occupation	race	sex	native_country	hours_per_week
0	0.030670	-1.884571	-0.991569	-0.866068	-1.378100	-0.353403	-0.703061	-0.255743	-0.035429
1	0.837096	-1.068730	-0.991569	-0.066951	-1.082777	-0.353403	-0.703061	-0.255743	-2.222119
2	-0.042641	-0.252888	-0.702015	0.732166	-0.787453	-0.353403	-0.703061	-0.255743	-0.035429
3	1.057031	-0.252888	-0.412460	-0.066951	-0.787453	1.240608	-0.703061	-0.255743	-0.035429
4	-0.775756	-0.252888	-0.991569	-0.066951	-0.492130	1.240608	1.422309	-0.057541	-0.035429

数据预测：

X_train,X_test,y_train,y_test = train_test_split(X3,y,test_size = 0.2)
knn = KNeighborsClassifier(n_neighbors=15,weights='distance')

knn.fit(X_train,y_train)

y_ = knn.predict(X_test)

result = y_ == y_test

result.mean()

0.8106863196683556

---

自带方法：

from sklearn.preprocessing import StandardScaler

s = StandardScaler()
X5 = s.fit_transform(X)
X5[:5]

array([[ 0.03067056, -1.88460023, -0.99158435, -0.8660817 , -1.37812112,
        -0.35340882, -0.70307135, -0.25574647, -0.03542945],
       [ 0.83710898, -1.0687461 , -0.99158435, -0.06695205, -1.08279326,
        -0.35340882, -0.70307135, -0.25574647, -2.22215312],
       [-0.04264203, -0.25289198, -0.70202542,  0.7321776 , -0.78746539,
        -0.35340882, -0.70307135, -0.25574647, -0.03542945],
       [ 1.05704673, -0.25289198, -0.4124665 , -0.06695205, -0.78746539,
         1.240627  , -0.70307135, -0.25574647, -0.03542945],
       [-0.77576787, -0.25289198, -0.99158435, -0.06695205, -0.49213753,
         1.240627  ,  1.42233076, -0.05754204, -0.03542945]])

5、保存模型与调用

5.1、保存模型

from sklearn.externals import joblib
joblib.dump(knn,'./model')

['./model']

5.2、加载模型

model = joblib.load('./model')
model

5.3、使用预测

model.score(X_test,y_test)

0.8174420389989252

保存为其他格式：