zip.train: Handwritten Digit Recognition Data

In ElemStatLearn: Data Sets, Functions and Examples from the Book: "The Elements of Statistical Learning, Data Mining, Inference, and Prediction" by Trevor Hastie, Robert Tibshirani and Jerome Friedman

Description

This example is a character recognition task: classification of handwritten numerals. This problem captured the attention of the machine learning and neural network community for many years, and has remained a benchmark problem in the field.

Usage

data(zip.train)

Format

The format is: num [1:7291, 1:257] 6 5 4 7 3 6 3 1 0 1 ...

Details

Normalized handwritten digits, automatically scanned from envelopes by the U.S. Postal Service. The original scanned digits are binary and of different sizes and orientations; the images here have been deslanted and size normalized, resulting in 16 x 16 grayscale images (Le Cun et al., 1990).

The data are in two gzipped files, and each line consists of the digit id (0-9) followed by the 256 grayscale values.

There are 7291 training observations and 2007 test observations, distributed as follows: 0 1 2 3 4 5 6 7 8 9 Total Train 1194 1005 731 658 652 556 664 645 542 644 7291 Test 359 264 198 166 200 160 170 147 166 177 2007

or as proportions: 0 1 2 3 4 5 6 7 8 9 Train 0.16 0.14 0.1 0.09 0.09 0.08 0.09 0.09 0.07 0.09 Test 0.18 0.13 0.1 0.08 0.10 0.08 0.08 0.07 0.08 0.09

The test set is notoriously "difficult", and a 2.5 excellent. These data were kindly made available by the neural network group at AT&T research labs (thanks to Yann Le Cunn).

Examples

findRows <- function(zip, n) {
 # Find  n (random) rows with zip representing 0,1,2,...,9
 res <- vector(length=10, mode="list")
 names(res) <- 0:9
 ind <- zip[,1]
 for (j in 0:9) {
    res[[j+1]] <- sample( which(ind==j), n ) }
 return(res) }

# Making a plot like that on page 4:

digits <- vector(length=10, mode="list")
names(digits) <- 0:9
rows <- findRows(zip.train, 6)
for (j in 0:9) {
    digits[[j+1]] <- do.call("cbind", lapply(as.list(rows[[j+1]]), 
                       function(x) zip2image(zip.train, x)) )
}
im <- do.call("rbind", digits)
image(im, col=gray(256:0/256), zlim=c(0,1), xlab="", ylab="" )    

Example output

[1] "digit  0  taken"
[1] "digit  0  taken"
[1] "digit  0  taken"
[1] "digit  0  taken"
[1] "digit  0  taken"
[1] "digit  0  taken"
[1] "digit  1  taken"
[1] "digit  1  taken"
[1] "digit  1  taken"
[1] "digit  1  taken"
[1] "digit  1  taken"
[1] "digit  1  taken"
[1] "digit  2  taken"
[1] "digit  2  taken"
[1] "digit  2  taken"
[1] "digit  2  taken"
[1] "digit  2  taken"
[1] "digit  2  taken"
[1] "digit  3  taken"
[1] "digit  3  taken"
[1] "digit  3  taken"
[1] "digit  3  taken"
[1] "digit  3  taken"
[1] "digit  3  taken"
[1] "digit  4  taken"
[1] "digit  4  taken"
[1] "digit  4  taken"
[1] "digit  4  taken"
[1] "digit  4  taken"
[1] "digit  4  taken"
[1] "digit  5  taken"
[1] "digit  5  taken"
[1] "digit  5  taken"
[1] "digit  5  taken"
[1] "digit  5  taken"
[1] "digit  5  taken"
[1] "digit  6  taken"
[1] "digit  6  taken"
[1] "digit  6  taken"
[1] "digit  6  taken"
[1] "digit  6  taken"
[1] "digit  6  taken"
[1] "digit  7  taken"
[1] "digit  7  taken"
[1] "digit  7  taken"
[1] "digit  7  taken"
[1] "digit  7  taken"
[1] "digit  7  taken"
[1] "digit  8  taken"
[1] "digit  8  taken"
[1] "digit  8  taken"
[1] "digit  8  taken"
[1] "digit  8  taken"
[1] "digit  8  taken"
[1] "digit  9  taken"
[1] "digit  9  taken"
[1] "digit  9  taken"
[1] "digit  9  taken"
[1] "digit  9  taken"
[1] "digit  9  taken"

ElemStatLearn documentation built on Aug. 12, 2019, 9:04 a.m.