MonadicContextualClassification | Wolfram Resource Object

Classification of MNIST data

Introduction	Data
Setup	Pipeline

Introduction

In this notebook we show how to make and run a Machine Learning (ML) classification pipeline over

MNIST data

of handwritten digits.

Setup

Load the monad

In[17]:=

Needs["AntonAntonov`MonadicContextualClassification`"]

Data

Here we show an example of using ClCon with the reasonably large dataset of images MNIST, [

YL1

].

In[18]:=

mnistData=ExampleData[{"MachineLearning","MNIST"},"Data"];

In[23]:=

mnistData//Dimensions

Out[23]=

{70000}

Pipeline

In[21]:=

SeedRandom[3423];p=

ClConUnit

[mnistData]⟹

ClConSplitData

[0.7]⟹

ClConSummarizeData

⟹

ClConMakeClassifier

["NearestNeighbors"]⟹

ClConClassifierMeasurements

[{"Accuracy","ConfusionMatrixPlot"}]⟹

ClConEchoValue

;

Out[21]=

RandomGeneratorState

Method: ExtendedCA

State hash: 8226606351136977657



»

summaries:trainingData

1 column 1

	1
	1
	1
	1
	1
	1
(Other)	48990



1 column 1

Min	0
1st Qu	2
Median	4
Mean	4.45242
3rd Qu	7
Max	9

,testData

1 column 1

	1
	1
	1
	1
	1
	1
(Other)	20998



1 column 1

Min	0
1st Qu	2
Median	4
Mean	4.45244
3rd Qu	7
Max	9



»

value:Accuracy0.960246,ConfusionMatrixPlot



Here we plot the ROC curve for a specified digit:

In[24]:=

p⟹

ClConROCPlot

["ClassLabels"5];

»

ROC plot(s):5



RelatedGuides

▪

Classification pipeline functions

RelatedTechNotes

▪

Classification workflows monad

""