十二、八种常用机器学习分类算法实战---通过六个因素对等级进行预测

一、需求分析

根据之前做训练的模型，对不同等级的标准样卡进行测试
测试样本有48张，其中包括起球个数、起球总面积、起球最大面积、起球平均面积、对比度、光学体积六个指标，最终确定出织物的等级
数据集fiber.csv大致结构如下：
（数据集是我自己测试收集的，这里就不公开分享了，个人数据，理解万岁）

csv格式注意事项：
N,S,Max_s,Aver_s,C,V,Grade最后没有空格
27,111542.5,38299.5,4131.2,31.91,3559537.61,1(空格)1后面有个空格，要注意！！！

变量	含义
N	起球个数
S	起球总面积
Max_s	起球最大面积
Aver_s	起球平均面积
C	对比度
V	光学体积
Grade	最终评定等级

二、尝试多种方法去实现预测等级评定

1、导包

pip install scikit-learn 安装sklearn相关包

import numpy as np
import pandas as pd 
import matplotlib.pyplot as plt
%matplotlib inline
import seaborn as sns
 
from sklearn.preprocessing import LabelEncoder
from sklearn.model_selection import train_test_split
 
from sklearn.linear_model import LogisticRegression
from sklearn.neighbors import KNeighborsClassifier
from sklearn.naive_bayes import BernoulliNB
from sklearn.naive_bayes import GaussianNB
from sklearn.naive_bayes import MultinomialNB
from sklearn.svm import LinearSVC
from sklearn.svm import SVC
from sklearn.tree import DecisionTreeClassifier
 
from sklearn.metrics import confusion_matrix, classification_report, accuracy_score

2、读取展示数据集

fiber = pd.read_csv("./fiber.csv")
fiber.head(15)

print(fiber)
"""
     N         S    Max_s    Aver_s      C           V  Grade
0   27  111542.5  38299.5   4131.20  31.91  3559537.61      1
1   27  110579.5  31220.0   3186.63  31.28  2690869.73      1
......
47   9   33853.0   6329.0   3761.44  41.17  1393863.42      4
"""

3、划分数据集

最后一列为结果，其余的六个因素为自变量

X = fiber.drop(['Grade'], axis=1)
Y = fiber['Grade']

将数据集分为验证集和测试集两部分
random_state随机数种子，保证每次分割训练集和测试集都相同

X_train, X_test, y_train, y_test = train_test_split(X, Y, random_state=0)

查看下shape值
训练集36个，测试集12个，总共也就48个数据

print(X_test.shape) #(36, 6)
print(y_train.shape) #(36,)
print(X_test.shape) #(12, 6)

4、不同算法拟合

①K近邻算法，KNeighborsClassifier()

n_neighbors：选取最近的点的个数
通过这4个数据来对其他的数据进行拟合

knn = KNeighborsClassifier(n_neighbors=4)

对训练集进行训练拟合

knn.fit(X_train,y_train)

对测试集X_test进行预测，得到预测结果y_pred

y_pred = knn.predict(X_test)

预测结果y_pred和正确的答案y_test进行对比，求均值mean，看下正确率accuracy

accuracy = np.mean(y_pred==y_test)
print(accuracy)

也可以看下最终的得分

score = knn.score(X_test,y_test)
print(score)

随便生成一条数据对模型进行测试
16,18312.5,6614.5,2842.31,25.23,1147430.19,2
最终的等级为2

test = np.array([[16,18312.5,6614.5,2842.31,25.23,1147430.19]])
prediction = knn.predict(test)
print(prediction)
"""
[2]
"""

这是从训练集中抽取的，实际肯定不能这样干，这里只是在进行测试而已。

K近邻算法完整代码

import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.neighbors import KNeighborsClassifier
from sklearn.metrics import confusion_matrix, classification_report, accuracy_score


fiber = pd.read_csv("./fiber.csv")
# 划分自变量和因变量
X = fiber.drop(['Grade'], axis=1)
Y = fiber['Grade']
#划分数据集
X_train, X_test, y_train, y_test = train_test_split(X, Y, random_state=0)

knn = KNeighborsClassifier(n_neighbors=4)
knn.fit(X_train,y_train)
y_pred = knn.predict(X_test)#模型预测结果
accuracy = np.mean(y_pred==y_test)#准确度
score = knn.score(X_test,y_test)#得分
print(accuracy)
print(score)

#测试
test = np.array([[16,18312.5,6614.5,2842.31,25.23,1147430.19]])#随便找的一条数据
prediction = knn.predict(test)#带入数据，预测一下
print(prediction)

②逻辑回归算法，LogisticRegression()

实例化逻辑回归对象

lr = LogisticRegression()

将训练集传入，进行训练拟合

lr.fit(X_train,y_train)#模型拟合

对测试集X_test进行预测，得到预测结果y_pred

y_pred = lr.predict(X_test)#模型预测结果

预测结果y_pred和正确的答案y_test进行对比，求均值mean，看下正确率accuracy

accuracy = np.mean(y_pred==y_test)
print(accuracy)

也可以看下最终的得分

score = lr.score(X_test,y_test)
print(score)

随便生成一条数据对模型进行测试
20,44882.5,10563,5623.88,27.15,3053651.65,1
最终的等级为1

test = np.array([[20,44882.5,10563,5623.88,27.15,3053651.65]])#随便找的一条数据，正确等级为1
prediction = lr.predict(test)#带入数据，预测一下
print(prediction)
"""
[1]
"""

这是从训练集中抽取的，实际肯定不能这样干，这里只是在进行测试而已。

逻辑回归完整代码

import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression


fiber = pd.read_csv("./fiber.csv")
# 划分自变量和因变量
X = fiber.drop(['Grade'], axis=1)
Y = fiber['Grade']
#划分数据集
X_train, X_test, y_train, y_test = train_test_split(X, Y, random_state=0)

lr = LogisticRegression()
lr.fit(X_train,y_train)#模型拟合
y_pred = lr.predict(X_test)#模型预测结果
accuracy = np.mean(y_pred==y_test)#准确度
score = lr.score(X_test,y_test)#得分
print(accuracy)
print(score)

test = np.array([[20,44882.5,10563,5623.88,27.15,3053651.65]])#随便找的一条数据
prediction = lr.predict(test)#带入数据，预测一下
print(prediction)

③线性支持向量机 ，LinearSVC()

实例化线性支持向量机对象

lsvc = LinearSVC()

将训练集传入，进行训练拟合

lsvc.fit(X_train,y_train)#模型拟合

对测试集X_test进行预测，得到预测结果y_pred

y_pred = lsvc.predict(X_test)#模型预测结果

预测结果y_pred和正确的答案y_test进行对比，求均值mean，看下正确率accuracy

accuracy = np.mean(y_pred==y_test)
print(accuracy)

也可以看下最终的得分

score = lsvc.score(X_test,y_test)
print(score)

随便生成一条数据对模型进行测试
20,55997.5,17644.5,2799.88,8.58,480178.56,2
最终的等级为2

test = np.array([[20,55997.5,17644.5,2799.88,8.58,480178.56]])#随便找的一条数据
prediction = lsvc.predict(test)#带入数据，预测一下
print(prediction)
"""
[2]
"""

这是从训练集中抽取的，实际肯定不能这样干，这里只是在进行测试而已。

线性支持向量机完整代码

from sklearn.svm import LinearSVC
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split

fiber = pd.read_csv("./fiber.csv")
# 划分自变量和因变量
X = fiber.drop(['Grade'], axis=1)
Y = fiber['Grade']
#划分数据集
X_train, X_test, y_train, y_test = train_test_split(X, Y, random_state=0)

lsvc = LinearSVC()
lsvc.fit(X_train,y_train)#模型拟合
y_pred = lsvc.predict(X_test)#模型预测结果
accuracy = np.mean(y_pred==y_test)#准确度
score = lsvc.score(X_test,y_test)#得分
print(accuracy)
print(score)

test = np.array([[20,55997.5,17644.5,2799.88,8.58,480178.56]])#随便找的一条数据
prediction = lsvc.predict(test)#带入数据，预测一下
print(prediction)

④支持向量机，SVC()

实例化支持向量机对象

svc = SVC()

将训练集传入，进行训练拟合

svc.fit(X_train,y_train)#模型拟合

对测试集X_test进行预测，得到预测结果y_pred

y_pred = svc.predict(X_test)#模型预测结果

预测结果y_pred和正确的答案y_test进行对比，求均值mean，看下正确率accuracy

accuracy = np.mean(y_pred==y_test)
print(accuracy)

也可以看下最终的得分

score = svc.score(X_test,y_test)
print(score)

随便生成一条数据对模型进行测试
23,97215.5,22795.5,2613.09,29.72,1786141.62,1
最终的等级为1

test = np.array([[23,97215.5,22795.5,2613.09,29.72,1786141.62]])#随便找的一条数据
prediction = svc.predict(test)#带入数据，预测一下
print(prediction)
"""
[1]
"""

这是从训练集中抽取的，实际肯定不能这样干，这里只是在进行测试而已。

支持向量机完整代码

from sklearn.svm import SVC
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split

fiber = pd.read_csv("./fiber.csv")
# 划分自变量和因变量
X = fiber.drop(['Grade'], axis=1)
Y = fiber['Grade']
#划分数据集
X_train, X_test, y_train, y_test = train_test_split(X, Y, random_state=0)

svc = SVC(gamma='auto')
svc.fit(X_train,y_train)#模型拟合
y_pred = svc.predict(X_test)#模型预测结果
accuracy = np.mean(y_pred==y_test)#准确度
score = svc.score(X_test,y_test)#得分
print(accuracy)
print(score)

test = np.array([[23,97215.5,22795.5,2613.09,29.72,1786141.62]])#随便找的一条数据
prediction = svc.predict(test)#带入数据，预测一下
print(prediction)

⑤决策树，DecisionTreeClassifier()

发现没，前四个方法步骤几乎都是一样的，只不过实例化的对象不同，仅此而已，至此就不再赘述了。

随便生成一条数据对模型进行测试
11,99498,5369,9045.27,28.47,3827588.56,4
最终的等级为4

决策树完整代码

from sklearn.tree import DecisionTreeClassifier
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split

fiber = pd.read_csv("./fiber.csv")
# 划分自变量和因变量
X = fiber.drop(['Grade'], axis=1)
Y = fiber['Grade']
#划分数据集
X_train, X_test, y_train, y_test = train_test_split(X, Y, random_state=0)

dtc = DecisionTreeClassifier()
dtc.fit(X_train,y_train)#模型拟合
y_pred = dtc.predict(X_test)#模型预测结果
accuracy = np.mean(y_pred==y_test)#准确度
score = dtc.score(X_test,y_test)#得分
print(accuracy)
print(score)

test = np.array([[11,99498,5369,9045.27,28.47,3827588.56]])#随便找的一条数据
prediction = dtc.predict(test)#带入数据，预测一下
print(prediction)

⑥高斯贝叶斯，GaussianNB()

随便生成一条数据对模型进行测试
14,160712,3208,3681.25,36.31,1871275.09,3
最终的等级为3

高斯贝叶斯完整代码

from sklearn.naive_bayes import GaussianNB
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split

fiber = pd.read_csv("./fiber.csv")
# 划分自变量和因变量
X = fiber.drop(['Grade'], axis=1)
Y = fiber['Grade']
#划分数据集
X_train, X_test, y_train, y_test = train_test_split(X, Y, random_state=0)

gnb = GaussianNB()
gnb.fit(X_train,y_train)#模型拟合
y_pred = gnb.predict(X_test)#模型预测结果
accuracy = np.mean(y_pred==y_test)#准确度
score = gnb.score(X_test,y_test)#得分
print(accuracy)
print(score)

test = np.array([[14,160712,3208,3681.25,36.31,1871275.09]])#随便找的一条数据
prediction = gnb.predict(test)#带入数据，预测一下
print(prediction)

⑦伯努利贝叶斯，BernoulliNB()

随便生成一条数据对模型进行测试
18,57541.5,10455,2843.36,30.68,1570013.02,2
最终的等级为2

伯努利贝叶斯完整代码

from sklearn.naive_bayes import BernoulliNB
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split

fiber = pd.read_csv("./fiber.csv")
# 划分自变量和因变量
X = fiber.drop(['Grade'], axis=1)
Y = fiber['Grade']
#划分数据集
X_train, X_test, y_train, y_test = train_test_split(X, Y, random_state=0)

bnb = BernoulliNB()
bnb.fit(X_train,y_train)#模型拟合
y_pred = bnb.predict(X_test)#模型预测结果
accuracy = np.mean(y_pred==y_test)#准确度
score = bnb.score(X_test,y_test)#得分
print(accuracy)
print(score)

test = np.array([[18,57541.5,10455,2843.36,30.68,1570013.02]])#随便找的一条数据
prediction = bnb.predict(test)#带入数据，预测一下
print(prediction)

⑧多项式贝叶斯，MultinomialNB()

随便生成一条数据对模型进行测试
9,64794,5560,10682.94,38.99,3748367.45,4
最终的等级为4

多项式贝叶斯完整代码

from sklearn.naive_bayes import MultinomialNB
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split

fiber = pd.read_csv("./fiber.csv")
# 划分自变量和因变量
X = fiber.drop(['Grade'], axis=1)
Y = fiber['Grade']
#划分数据集
X_train, X_test, y_train, y_test = train_test_split(X, Y, random_state=0)

mnb = MultinomialNB()
mnb.fit(X_train,y_train)#模型拟合
y_pred = mnb.predict(X_test)#模型预测结果
accuracy = np.mean(y_pred==y_test)#准确度
score = mnb.score(X_test,y_test)#得分
print(accuracy)
print(score)

test = np.array([[9,64794,5560,10682.94,38.99,3748367.45]])#随便找的一条数据
prediction = mnb.predict(test)#带入数据，预测一下
print(prediction)

最终通过调节参数以及优化，确定使用决策树对本样本进行等级预测

5、模型保存与加载

这里以决策树算法为例

训练完成之后的模型通过joblib.dump(dtc, './dtc.model')进行保存
dtc为模型实例化对象
./dtc.model为保存模型名称和路径

通过dtc_yy = joblib.load('./dtc.model')加载模型

完整代码

from sklearn.tree import DecisionTreeClassifier
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
import joblib

fiber = pd.read_csv("./fiber.csv")
# 划分自变量和因变量
X = fiber.drop(['Grade'], axis=1)
Y = fiber['Grade']
#划分数据集
X_train, X_test, y_train, y_test = train_test_split(X, Y, random_state=0)

dtc = DecisionTreeClassifier()
dtc.fit(X_train,y_train)#模型拟合
joblib.dump(dtc, './dtc.model')#保存模型
y_pred = dtc.predict(X_test)#模型预测结果
accuracy = np.mean(y_pred==y_test)#准确度
score = dtc.score(X_test,y_test)#得分
print(accuracy)
print(score)


dtc_yy = joblib.load('./dtc.model')
test = np.array([[11,99498,5369,9045.27,28.47,3827588.56]])#随便找的一条数据
prediction = dtc_yy.predict(test)#带入数据，预测一下
print(prediction)

保存的模型如下：