训练多个分类器并比较指标

Question

我想一次计算不同的分类器并将结果传输到 Pandas 数据框。

# Lets create some test data
import pandas as pd
import numpy as np
import string 
import random
integers = pd.DataFrame(np.random.randint(0,100,size=(50, 1)), columns=list('I'))
strings = pd.DataFrame([random.choice('ab') for _ in range(50)], columns=list('S'))
df2 = pd.concat([strings,integers], axis=1)
df2.head()
    S   I
0   a   5
1   a   31
2   b   84
3   a   79
4   b   92


# Train - Test
from sklearn.model_selection import train_test_split

X = df2[["I"]].values
y = df2["S"]
X_train, X_test, y_train, y_test = train_test_split(X, y)

#Load libraries
from sklearn import metrics
from sklearn.model_selection import cross_val_score
from sklearn.neighbors import KNeighborsClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier, AdaBoostClassifier
from sklearn.linear_model import LogisticRegression


#Classifiers 
classifiers = [
KNeighborsClassifier(30),
DecisionTreeClassifier(),
RandomForestClassifier(),
AdaBoostClassifier(),
LogisticRegression()]


n_range = list(range(1, 10))
RandomForestClf = []
data_frame = []

for n in n_range:
#    name = clf.__class__.__name__
model = RandomForestClassifier(n_estimators=n)
scores = cross_val_score(model, X, y, cv=5, scoring="accuracy")
RandomForestClf.append(scores.mean())
data_frame = pd.DataFrame({"Random Forest": RandomForestClf})

我无法让各种分类器通过 for 循环。

如何设置 for 循环，以便运行每个分类器，然后将预测转移到 panda 数据帧？

我当前的 for 循环只有在代码中提到模型时才有效。

我是 Python 新手。

感谢您的帮助！

Answer 1

您可以在 for 循环之外定义数据框，然后分配给它查找分类器名称，检查对象的 type：

from sklearn import metrics
from sklearn.model_selection import cross_val_score
from sklearn.neighbors import KNeighborsClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier, AdaBoostClassifier
from sklearn.linear_model import LogisticRegression


#Classifiers 
classifiers = [KNeighborsClassifier(30),
                DecisionTreeClassifier(),
                RandomForestClassifier(),
                AdaBoostClassifier(),
                LogisticRegression()]

from sklearn.datasets import load_iris
X, y = load_iris(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y)

k = 5
preds = pd.DataFrame(index=[*range(k)])

for cls in classifiers:
    scores = cross_val_score(cls, X, y, cv=k, scoring="accuracy")
    preds[type(cls).__name__] = scores

在这种情况下，你会得到：

print(preds)
   KNeighborsClassifier  DecisionTreeClassifier  RandomForestClassifier  \
0              0.900000                0.966667                0.966667   
1              0.966667                0.966667                0.966667   
2              0.933333                0.900000                0.933333   
3              0.900000                0.966667                0.966667   
4              1.000000                1.000000                1.000000   

   AdaBoostClassifier  LogisticRegression  
0            0.966667            0.966667  
1            0.933333            1.000000  
2            0.900000            0.933333  
3            0.933333            0.966667  
4            1.000000            1.000000   

这是一个related answer，它从分类器列表中绘制多个混淆矩阵，以防您也发现它也很有用。