Pattern Recognition and Neural Networks by B.D. Ripley

This collection contains datasets used in the book Pattern Recognition and Neural Networks by B.D. Ripley (1996) Cambridge University Press ISBN 0 521 46986 7 (hardback, 416 pages, 29.95 pounds, US$49.95)

The background to the datasets is described in section 1.4; this file relates the computer-readable files to that description.

Cushing's syndrome

Data from Aitchison & Dunsmore (1975, Tables 11.1-3).

Data file Cushings.dat has four columns, The type of the last six patients (u1 to u6) should be regarded as unknown. (The code `o' indicates `other').

synthetic two-class problem

Data from Ripley (1994a).
This has two real-valued co-ordinates (xs and ys) and a class (xc) which is 0 or 1.
has 250 rows of the training set
has 1000 rows of the test set (not used here)


This is a dataset on 61 viruses with rod-shaped particles affecting various crops (tobacco, tomato, cucumber and others) described by {Fauquet et al. (1988) and analysed by Eslava-G\'omez (1989). There are 18 measurements on each virus, the number of amino acid residues per molecule of coat protein.

has 61 rows of 18 counts
has 38 rows corresponding to the distinct Tobamoviruses.
The whole dataset is in order Hordeviruses (3), Tobraviruses (6), Tobamoviruses (39) and `furoviruses' (13).

Leptograpsus crabs

Data from Campbell & Mahon (1974) on the morphology of rock crabs of genus Leptograpsus. There are 50 specimens of each sex of each of two colour forms.

Data file crabs.dat has rows

Forensic glass

This example comes from forensic testing of glass collected by B. German on 214 fragments of glass. It is also contained in the UCI machine-learning database collection (Murphy & Aha, 1995).

Data file fglass.dat has 214 rows with data for a single glass fragment. The type codes are:
  1. (WinF) window float glass
  2. (WinNF) window non-float glass
  3. (Veh) vehicle glass
  4. (not used)
  5. (Con) containers
  6. (Tabl) tableware
  7. (Head) vehicle headlamp glass
The ten groups used for the cross-validation experiments (I believe) are listed as row numbers in the file fglass.grp

Diabetes in Pima Indians

A population of women who were at least 21 years old, of Pima Indian heritage and living near Phoenix, Arizona, was tested for diabetes according to World Health Organization criteria. The data were collected by the US National Institute of Diabetes and Digestive and Kidney Diseases (Smith et al, 1988). This example is also contained in the UCI machine-learning database collection (Murphy & Aha, 1995).

The data files have rows containing
npreg -- number of pregnancies
glu -- plasma glucose concentration in an oral glucose tolerance test
bp -- diastolic blood pressure (mm Hg)
skin -- triceps skin fold thickness (mm)
ins -- serum insulin (micro U/ml)
bmi -- body mass index (weight in kg/(height in m)^2)
ped -- diabetes pedigree function
age -- in years
type -- No / Yes
has 200 rows of complete training data.
has 332 rows of complete test data.
has the 200 rows of plus 100 incomplete rows.
Last edited on Tues Nov 7 1995 by Brian Ripley (