Classify
ClassifierMeasurements
ExampleData
Dataset
AssociationMap

Machine Learning
Data Science

Visualize the Confusion Matrix of a Classifier

Open in Cloud
Download Notebook

Classify the type of a satellite image as a function of features of the image

Collect examples from a built-in dataset:

In[1]:=

trainingset=ExampleData[{"MachineLearning","Satellite"},"TrainingData"];testset=ExampleData[{"MachineLearning","Satellite"},"TestData"];

Train a classifier on the training set:

In[2]:=

c=Classify[trainingset]

Out[2]=

ClassifierFunction



Input type:

Mixed

(number: 36)

Number of classes:

Data not in notebook. Store now



Get data on the performance of the classifier on the test set:

In[3]:=

cm=ClassifierMeasurements[c,testset];

The measurement object can be queried for properties such as the accuracy of the classifier:

In[4]:=

cm["Properties"]//Short

Out[4]//Short=

{Accuracy,AccuracyBaseline,AccuracyRejectionPlot,53,TruePositiveNumber,WorstClassifiedExamples}

In[5]:=

cm["Accuracy"]

Out[5]=

0.8915

In[6]:=

AssociationMap[cm,{"Precision","Recall","F1Score"}]//Dataset//Transpose

Out[6]=

Compute the confusion matrix either in numeric form or as annotated graphics:

In[7]:=

cm["ConfusionMatrix"]//MatrixForm

Out[7]//MatrixForm=

217	0	1	0	4	2
1	135	30	0	2	43
1	15	368	3	0	10
1	0	1	458	1	0
4	2	1	15	191	24
0	30	16	0	10	414

In[8]:=

cm["ConfusionMatrixPlot"]

Out[8]=

Related Symbols

Classify
ClassifierMeasurements
ExampleData
Dataset
AssociationMap

Publisher Information

Contributed by: Wolfram Staff

Wolfram Language Example Repository

Visualize the Confusion Matrix of a Classifier

See Also

Publisher Information

217	0	1	0	4	2
1	135	30	0	2	43
1	15	368	3	0	10
1	0	1	458	1	0
4	2	1	15	191	24
0	30	16	0	10	414

217	0	1	0	4	2
1	135	30	0	2	43
1	15	368	3	0	10
1	0	1	458	1	0
4	2	1	15	191	24
0	30	16	0	10	414

217	0	1	0	4	2
1	135	30	0	2	43
1	15	368	3	0	10
1	0	1	458	1	0
4	2	1	15	191	24
0	30	16	0	10	414