Python機器學習: train_test_split() 切割資料(波士頓地區房價)為訓練資料跟測試資料; from sklearn.model_selection import train_test_split ; xtrain, xtest, ytrain, ytest = train_test_split(x, y, test_size=0.3, random_state=42, shuffle=True)

from sklearn.model_selection import train_test_split

xtrain,xtest,ytrain,ytest =\
train_test_split(x,y,test_size=0.3,
random_state=42,shuffle=True)

樣品分割:

參考解答:

import pandas as pd
import sys
fpath = r”C:\Python\P107\BostonHousing.csv”
dataset = pd.read_csv(fpath)
#.shape = (506, 14)
headerList = dataset.columns.tolist()
cols = dataset.columns.size
# dataset.columns.size = 14
x = dataset.drop(headerList[-1],axis=1).values
y = dataset[headerList[-1]].values

“””

#x=dataset.iloc[:,0:cols-1].values
#y=dataset.iloc[:,cols-1:].values

#用iloc切片的y 資料非常像,

#但變成2D,第二維長度僅有1

#y=dataset.iloc[:,cols-1:].values.ravel()

#用ravel() 把2D降維成1D

y剛好在最後一欄比較好切片

若沒在最後一欄,

原本的drop比較好用

“””

from sklearn.model_selection import train_test_split
xtrain,xtest,ytrain,ytest =\
train_test_split(x,y,test_size=0.3,
random_state=42,shuffle=True)
print(“The shpae of training data(X axis):”,xtrain.shape)
print(“The shpae of training data(Y axis):”,ytrain.shape)
print(“The shpae of testing data(X axis):”,xtest.shape)
print(“The shpae of testing data(Y axis):”,ytest.shape)

506*0.3 = 152 (test_size=0.3)

506-152 = 354

改用iloc做切片

.ravel() 將2D降維成1D

實測random_state跟shuffle參數:

若真的需要都次取的都不一樣

random_state = None

shuffle = True

其他組合每次都會

取出一樣的samples

每次都一樣,

會比較容易debug

或判斷模型的優劣

random_state設了一樣的種子

shuffle = True也無效

每次都會取出一樣的資料

第四個組合

random_state = None

shuffle = False

沒種子,沒洗牌

為什麼每次都一樣?

from官網:

波士頓地區房價:

boston (整齊版):

boston(排版較亂): http://lib.stat.cmu.edu/datasets/boston

要如以下才能拼接回資料

import pandas as pd
import numpy as np
url = r”http://lib.stat.cmu.edu/datasets/boston”
dfBoston = pd.read_csv(url,sep= “\s+” ,skiprows=22,header=None)

#正則表示法, 一個以上的空白或Tab
x=np.hstack( [ dfBoston.values[::2,:],dfBoston.values[1::2,:2] ] )

y=dfBoston.values[1::2,2]

部分y:

y.shape