source: orange/orange/doc/datasets/wine.htm @ 1760:9d4bb141fb0e

Revision 1760:9d4bb141fb0e, 2.7 KB checked in by blaz <blaz.zupan@…>, 9 years ago (diff)

data info file

Line 
1<html>
2<head>
3<title>Wine Recognition Data Base</title>
4</head>
5<body>
6<h1>Info on Wine Recognition Data Base</h1>
7<pre>
81. Title of Database: Wine recognition data
92. Sources:
10   (a) Forina, M. et al, PARVUS - An Extendible Package for Data
11       Exploration, Classification and Correlation. Institute of Pharmaceutical
12       and Food Analysis and Technologies, Via Brigata Salerno,
13       16147 Genoa, Italy.
14
15   (b) Stefan Aeberhard, email: stefan@coral.cs.jcu.edu.au
16   (c) July 1991
173. Past Usage:
18
19   (1)
20   S. Aeberhard, D. Coomans and O. de Vel,
21   Comparison of Classifiers in High Dimensional Settings,
22   Tech. Rep. no. 92-02, (1992), Dept. of Computer Science and Dept. of
23   Mathematics and Statistics, James Cook University of North Queensland.
24   (Also submitted to Technometrics).
25
26   The data was used with many others for comparing various
27   classifiers. The classes are separable, though only RDA
28   has achieved 100% correct classification.
29   (RDA : 100%, QDA 99.4%, LDA 98.9%, 1NN 96.1% (z-transformed data))
30   (All results using the leave-one-out technique)
31
32   In a classification context, this is a well posed problem
33   with "well behaved" class structures. A good data set
34   for first testing of a new classifier, but not very
35   challenging.
36
37   (2)
38   S. Aeberhard, D. Coomans and O. de Vel,
39   "THE CLASSIFICATION PERFORMANCE OF RDA"
40   Tech. Rep. no. 92-01, (1992), Dept. of Computer Science and Dept. of
41   Mathematics and Statistics, James Cook University of North Queensland.
42   (Also submitted to Journal of Chemometrics).
43
44   Here, the data was used to illustrate the superior performance of
45   the use of a new appreciation function with RDA.
46
474. Relevant Information:
48
49   -- These data are the results of a chemical analysis of
50      wines grown in the same region in Italy but derived from three
51      different cultivars.
52      The analysis determined the quantities of 13 constituents
53      found in each of the three types of wines.
54
55   -- I think that the initial data set had around 30 variables, but
56      for some reason I only have the 13 dimensional version.
57      I had a list of what the 30 or so variables were, but a.)
58      I lost it, and b.), I would not know which 13 variables
59      are included in the set.
60
615. Number of Instances
62
63        class 1 59
64    class 2 71
65    class 3 48
66
676. Number of Attributes
68   
69    13
70
717. For Each Attribute:
72
73    All attributes are continuous
74   
75    No statistics available, but suggest to standardise
76    variables for certain uses (e.g. for us with classifiers
77    which are NOT scale invariant)
78
79    NOTE: 1st attribute is class identifier (1-3)
80
818. Missing Attribute Values:
82
83    None
84
859. Class Distribution: number of instances per class
86
87        class 1 59
88    class 2 71
89    class 3 48
90</pre>
91</body>
92</html>
Note: See TracBrowser for help on using the repository browser.