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

Revision 1760:9d4bb141fb0e, 3.6 KB checked in by blaz <blaz.zupan@…>, 9 years ago (diff)

data info file

Line 
1<html>
2<head>
3<title>Glass Indetification Data Base</title>
4</head>
5<body>
6<h1>Info on Glass Identification Data Base</h1>
7<pre>
81. Title: Glass Identification Database
9
102. Sources:
11    (a) Creator: B. German
12        -- Central Research Establishment
13           Home Office Forensic Science Service
14           Aldermaston, Reading, Berkshire RG7 4PN
15    (b) Donor: Vina Spiehler, Ph.D., DABFT
16               Diagnostic Products Corporation
17               (213) 776-0180 (ext 3014)
18    (c) Date: September, 1987
19
203. Past Usage:
21    -- Rule Induction in Forensic Science
22       -- Ian W. Evett and Ernest J. Spiehler
23       -- Central Research Establishment
24          Home Office Forensic Science Service
25          Aldermaston, Reading, Berkshire RG7 4PN
26       -- Unknown technical note number (sorry, not listed here)
27       -- General Results: nearest neighbor held its own with respect to the
28             rule-based system
29
304. Relevant Information:n
31      Vina conducted a comparison test of her rule-based system, BEAGLE, the
32      nearest-neighbor algorithm, and discriminant analysis.  BEAGLE is
33      a product available through VRS Consulting, Inc.; 4676 Admiralty Way,
34      Suite 206; Marina Del Ray, CA 90292 (213) 827-7890 and FAX: -3189.
35      In determining whether the glass was a type of "float" glass or not,
36      the following results were obtained (# incorrect answers):
37
38             Type of Sample                            Beagle   NN    DA
39             Windows that were float processed (87)     10      12    21
40             Windows that were not:            (76)     19      16    22
41
42      The study of classification of types of glass was motivated by
43      criminological investigation.  At the scene of the crime, the glass left
44      can be used as evidence...if it is correctly identified!
45
465. Number of Instances: 214
47
486. Number of Attributes: 10 (including an Id#) plus the class attribute
49   -- all attributes are continuously valued
50
517. Attribute Information:
52   1. Id number: 1 to 214
53   2. RI: refractive index
54   3. Na: Sodium (unit measurement: weight percent in corresponding oxide, as
55                  are attributes 4-10)
56   4. Mg: Magnesium
57   5. Al: Aluminum
58   6. Si: Silicon
59   7. K: Potassium
60   8. Ca: Calcium
61   9. Ba: Barium
62  10. Fe: Iron
63  11. Type of glass: (class attribute)
64      -- 1 building_windows_float_processed
65      -- 2 building_windows_non_float_processed
66      -- 3 vehicle_windows_float_processed
67      -- 4 vehicle_windows_non_float_processed (none in this database)
68      -- 5 containers
69      -- 6 tableware
70      -- 7 headlamps
71
728. Missing Attribute Values: None
73
74Summary Statistics:
75Attribute:   Min     Max      Mean     SD      Correlation with class
76 2. RI:       1.5112  1.5339   1.5184  0.0030  -0.1642
77 3. Na:      10.73   17.38    13.4079  0.8166   0.5030
78 4. Mg:       0       4.49     2.6845  1.4424  -0.7447
79 5. Al:       0.29    3.5      1.4449  0.4993   0.5988
80 6. Si:      69.81   75.41    72.6509  0.7745   0.1515
81 7. K:        0       6.21     0.4971  0.6522  -0.0100
82 8. Ca:       5.43   16.19     8.9570  1.4232   0.0007
83 9. Ba:       0       3.15     0.1750  0.4972   0.5751
8410. Fe:       0       0.51     0.0570  0.0974  -0.1879
85
869. Class Distribution: (out of 214 total instances)
87    -- 163 Window glass (building windows and vehicle windows)
88       -- 87 float processed 
89          -- 70 building windows
90          -- 17 vehicle windows
91       -- 76 non-float processed
92          -- 76 building windows
93          -- 0 vehicle windows
94    -- 51 Non-window glass
95       -- 13 containers
96       -- 9 tableware
97       -- 29 headlamps
98</pre>
99</body>
100</html>
Note: See TracBrowser for help on using the repository browser.