Wolfram Data Repository

Immediate Computable Access to Curated Contributed Data

Sample Data: UCI Letter

Source Notebook

Letter recognition dataset

This dataset originates from a letter recognition task, where images of the 26 capital letters in the English alphabet should be correctly classified given 16 numerical features of the image.

The example letters were obtained by generating letters using 20 different fonts, which were then randomly distorted to produce 20,000 unique images. Sixteen features (such as pixel counts and correlations) were then extracted from each image.

This dataset comes with predefined test and training sets. The test set consists of the first 16,000 examples with the rest used for training.

Examples

Basic Examples

Retrieve the resource:

In[1]:=
ResourceObject["Sample Data: UCI Letter"]
Out[1]=

Retrieve the default content:

In[2]:=
RandomSample[ResourceData["Sample Data: UCI Letter"], 5]
Out[2]=

Analysis

Train a classifier:

In[3]:=
c = Classify[ResourceData["Sample Data: UCI Letter", "TrainingData"]]
Out[6]=

Obtain general information about the classifier:

In[7]:=
Information[c]
Out[7]=

Visualize the accuracy of the classifier:

In[8]:=
ClassifierMeasurements[c, ResourceData["Sample Data: UCI Letter", "TestData"]]["ConfusionMatrixPlot"]
Out[9]=

Wolfram Research, "Sample Data: UCI Letter" from the Wolfram Data Repository (2019)  

Data Resource History

  • Date Created:

Source Metadata

  • Publisher: Dua, D. and Graff, C. (2019). UCI Machine Learning Repository [http://archive.ics.uci.edu/ml]. Irvine, CA: University of California, School of Information and Computer Science.
  • Citation: P. W. Frey and D. J. Slate (1991) Machine Learning Vol 6 #2.

Data Downloads

Publisher Information

  • Publisher of Record: Wolfram Research