2.2 Logistic Regression

2.2.1 Iris Data Set

In[]:=
session=StartExternalSession["Python"]
Out[]=
ExternalSessionObject
System: Python
EvaluationCount:
None
UUID: e4cd4b26-b7f7-44c2-a52b-f77a434e81a3

Test
〉
In[]:=
5+6
Out[]=
11
〉
In[]:=
from sklearn import datasets
iris=datasets.load_iris()
〉
In[]:=
X=iris.data[:,[2,3]]
y=iris.target
〉
In[]:=
import numpy as np
〉
In[]:=
np.savetxt('D:\\dataX.txt',X,fmt='%.2e')
〉
In[]:=
from sklearn.linear_model import LogisticRegression
lr=LogisticRegression(C=100.0,random_state=1).fit(X,y)
〉
In[]:=
lr.predict(X[:150,:])
Out[]=
NumericArray
Type: Integer32
Dimensions: {150}

In[]:=
yP=Normal[%]
Out[]=
{0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,2,1,1,1,1,1,1,2,1,1,1,1,1,2,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,1,2,2,2,2,2,2,2,2,2,2,2,2,2,1,1,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2}
〉
In[]:=
y
Out[]=
NumericArray
Type: Integer32
Dimensions: {150}

In[]:=
yTR=Normal[%]
Out[]=
{0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2}
In[]:=
Norm[yP-yTR]
Out[]=
6
〉
In[]:=
print("Training set score: {:.2f}".format(lr.score(X, y)))
Training set score: 0.96
In[]:=
Xtrain=Import["D:\\dataX.txt","Table"];
In[]:=
dataT=MapThread[Join[#1,{#2}]&,{Xtrain,yTR}];
In[]:=
data0=Map[{#[[1]],#[[2]]}&,Select[dataT,#[[3]]0&]];​​data1=Map[{#[[1]],#[[2]]}&,Select[dataT,#[[3]]1&]];data2=Map[{#[[1]],#[[2]]}&,Select[dataT,#[[3]]2&]];
In[]:=
p0=ListPlot[{data0,data1,data2},PlotStyle{Green,Blue,Red},FrameTrue,AxesNone,PlotMarkers{Automatic},AspectRatio0.9,FrameLabel{"petallength","petal width"},FrameTrue]​​
Out[]=
In[]:=
dataTrain=Map[{#[[1]],#[[2]]}#[[3]]&,dataT];​​
In[]:=
c=Classify[dataTrain,Method"LogisticRegression",PerformanceGoal"Quality"];​​
Then the accuracy of the classifier on the training set
In[]:=
ctraining=ClassifierMeasurements[c,dataTrain]​​
Out[]=
ClassifierMeasurementsObject
Classifier: LogisticRegression
Number of test examples: 150

In[]:=
ctraining["Accuracy"]
Out[]=
1.
In[]:=
ctraining["ConfusionMatrixPlot"]​​
Out[]=
In[]:=
Show[{DensityPlot[c[{u,v}],{u,0,7},{v,-1,4.5},ColorFunction"CMYKColors",PlotPoints50],p0},AspectRatio1.2]​​
Out[]=
In[]:=
yP=Map[c[#]&,Join[{data0,data1,data2}]]//Flatten;​​
In[]:=
Norm[yTR-yP]​​
Out[]=
6

2.2.2 Digit Recognition

〉
In[]:=
import numpy as np
​
​from numpy import array, matrix
from scipy.io import mmread, mmwrite
〉
In[]:=
X=mmread('Xtrain.mtx')
y=mmread('ytrain.mtx')
yy=y[0]
〉
In[]:=
from sklearn.linear_model import LogisticRegression
lr=LogisticRegression(C=100.,random_state=1).fit(X,yy)
〉
In[]:=
ytr=lr.predict(X)
〉
In[]:=
ytr
〉
In[]:=
yy
〉
In[]:=
print("Training set score: {:.2f}".format(lr.score(X, yy)))
〉
In[]:=
Xt=mmread('Xtest.mtx')
yt=mmread('ytest.mtx')
yt
〉
In[]:=
yte=lr.predict(Xt)
〉
In[]:=
yte
〉
In[]:=
print("Test set score: {:.2f}".format(lr.score(Xt, yt[0])))