main main
KEEL-dataset - data set description

This section describes main characteristics of the optdigits data set and its attributes:

General information

Optical Recognition of Handwritten Digits data set
TypeClassificationOriginReal world
Features 64(Real / Integer / Nominal)(0 / 64 / 0)
Instances5620 Classes10
Missing values?No

Attribute description

Atr-0[0, 0]Atr-22[0, 16]Atr-43[0, 16]
Atr-1[0, 8]Atr-23[0, 8]Atr-44[0, 16]
Atr-2[0, 16]Atr-24[0, 1]Atr-45[0, 16]
Atr-3[0, 16]Atr-25[0, 16]Atr-46[0, 16]
Atr-4[0, 16]Atr-26[0, 16]Atr-47[0, 6]
Atr-5[0, 16]Atr-27[0, 16]Atr-48[0, 10]
Atr-6[0, 16]Atr-28[0, 16]Atr-49[0, 16]
Atr-7[0, 16]Atr-29[0, 16]Atr-50[0, 16]
Atr-8[0, 5]Atr-30[0, 16]Atr-51[0, 16]
Atr-9[0, 16]Atr-31[0, 2]Atr-52[0, 16]
Atr-10[0, 16]Atr-32[0, 1]Atr-53[0, 16]
Atr-11[0, 16]Atr-33[0, 15]Atr-54[0, 16]
Atr-12[0, 16]Atr-34[0, 16]Atr-55[0, 13]
Atr-13[0, 16]Atr-35[0, 16]Atr-56[0, 1]
Atr-14[0, 16]Atr-36[0, 16]Atr-57[0, 10]
Atr-15[0, 15]Atr-37[0, 16]Atr-58[0, 16]
Atr-16[0, 5]Atr-38[0, 14]Atr-59[0, 16]
Atr-17[0, 16]Atr-39[0, 0]Atr-60[0, 16]
Atr-18[0, 16]Atr-40[0, 7]Atr-61[0, 16]
Atr-19[0, 16]Atr-41[0, 16]Atr-62[0, 16]
Atr-20[0, 16]Atr-42[0, 16]Atr-63[0, 16]
Atr-21[0, 16]Output{0, ..., 9}

Additional information

A preprocessing programs made available by NIST was used to extract normalized bitmaps of handwritten digits from a preprinted form. 32x32 bitmaps are divided into nonoverlapping blocks of 4x4 and the number of on pixels are counted in each block. This generates an input matrix of 8x8 where each element is an integer in the range 0..16. This reduces dimensionality and gives invariance to small distortions.

In this section you can download some files related to the optdigits data set:

  • The complete data set already formatted in KEEL format can be downloaded from herezip.gif.
  • A copy of the data set already partitioned by means of a 10-folds cross validation procedure can be downloaded from herezip.gif.
  • A copy of the data set already partitioned by means of a 5-folds cross validation procedure can be downloaded from herezip.gif.
  • The header file associated to this data set can be downloaded from heretxt.png.
  • This is not a native data set from the KEEL project. It has been obtained from the UCI Machine Learning Repository. The original page where the data set can be found is:

 Copyright 2004-2018, KEEL (Knowledge Extraction based on Evolutionary Learning)
About the Webmaster Team
Valid XHTML 1.1   Valid CSS!