Multiple Features Data Set of Robert P.W. Duin.

Share:

Description

This dataset consists of features of handwritten numerals (‘0’–‘9’) (K=10) extracted from a collection of Dutch utility maps. Two hundred patterns per class (for a total of 2,000 (=N) patterns) have been digitized in binary images. Raw observations are 32x45 bitmmaps, which are divided into nooverlapping blocks of 2x3 and the number of pixels are counted in each block. This generate p=240 (16x15) variable, recodring the normalized counts of pixels in each block and each element is an integer in the range 0 to 6. rownames of DutchUtility contains the true digits and colnames of it contains the position of the block matrix, from which the normalized counts of pixels are taken.

Usage

1
2
data(DutchUtility)
showDigit(index,cex.main=1)

Arguments

index

A scalar containing integers between 1 and 2000. The function ShowDigit regenerates the sampled versions of the original images may be obtained (15x16 pixels). (the source image (32x45) dataset is lost)

cex.main

Specify the size of the title text with a numeric value of length 1.

Details

The original dataset is freely available from USIMachine Learning Repository (Frank and Asuncion (2010)) website http://archive .ics.uci.edu/ml/datasets.html.

Author(s)

Yumi Kondo <y.kondo@stat.ubc.ca>

References

Frank A, Asuncion A (2010). UCI Machine Learning Repository." http://archive.ics.uci.edu/ml.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
## Not run: 

data(DutchUtility)

truedigit <- rownames(DutchUtility)
(re <- RSKC(DutchUtility,ncl=10,alpha=0.1,L1=5.7,nstart=1000))
Sensitivity(re$labels,truedigit)
table(re$labels,truedigit)

## Check the bitmap of the trimmed observations 
showDigit(re$oW[1])
## Check the features which receive zero weights
names(which(re$weights==0))

## End(Not run)