1 | .EN |
---|
2 | .TH C4.5 1 |
---|
3 | .SH NAME |
---|
4 | .PP |
---|
5 | consult \- classify items using a decision tree |
---|
6 | .SH SYNOPSIS |
---|
7 | .PP |
---|
8 | .B consult |
---|
9 | [ \fB-f\fR FNS ] |
---|
10 | [ \fB-t\fR ] |
---|
11 | .SH DESCRIPTION |
---|
12 | .PP |
---|
13 | .I Consult |
---|
14 | reads a decision tree produced by c4.5 and |
---|
15 | uses this to classify items whose description |
---|
16 | is provided by the user. |
---|
17 | The decision tree read is |
---|
18 | .I filestem.tree, |
---|
19 | where |
---|
20 | .I filestem |
---|
21 | is the file name stem (see manual entry for c4.5). |
---|
22 | .PP |
---|
23 | .I Consult |
---|
24 | prompts for the value of an attribute when needed. |
---|
25 | Unknown attribute values are specified as '?'. |
---|
26 | .PP |
---|
27 | If the attribute has discrete values, its value |
---|
28 | can be specified either as |
---|
29 | .ti 8 |
---|
30 | .I v |
---|
31 | .br |
---|
32 | where v is one of the permissible values of the attribute, or as |
---|
33 | .ti 8 |
---|
34 | .I "v1:p1, v2:p2, ... , vn:pn" |
---|
35 | .br |
---|
36 | where the vi's are some or all of the permissible values of the |
---|
37 | attribute and the pi's are corresponding probabilities. |
---|
38 | If the sum of the pi's is less than 1, the remaining probability |
---|
39 | is distributed equally among the unspecified values of the |
---|
40 | attribute. For example, if attribute A has possible values |
---|
41 | X, Y and Z, the reply |
---|
42 | .ti 8 |
---|
43 | X:0.5 |
---|
44 | .br |
---|
45 | is the same as |
---|
46 | .ti 8 |
---|
47 | X:0.5, Y:0.25, Z:0.25 |
---|
48 | .PP |
---|
49 | If the attribute has real values, its value can be |
---|
50 | specified either as |
---|
51 | .ti 8 |
---|
52 | .I r |
---|
53 | .br |
---|
54 | where r is a real number, or as |
---|
55 | .ti 8 |
---|
56 | .I r1 - r2 |
---|
57 | .br |
---|
58 | where r1 and r2 are real numbers. The latter form is |
---|
59 | taken to mean "somewhere in the interval [r1,r2]". |
---|
60 | .PP |
---|
61 | To display the subtree of the decision tree with the |
---|
62 | current node at the root, '?t' may be entered in |
---|
63 | response to a request for an attribute value. |
---|
64 | .PP |
---|
65 | When all relevant attributes have been determined, |
---|
66 | consult will give one or more classes that the item |
---|
67 | might belong to. The likelihood of a class may be |
---|
68 | indicated by a probability, followed sometimes by a |
---|
69 | probability interval. For example, |
---|
70 | .ti 8 |
---|
71 | .I "C1 CF = 0.9 [0.85 - 1]" |
---|
72 | .br |
---|
73 | means "class C1 with probability in the interval 0.85 - 1, |
---|
74 | and with best guess probability 0.9". |
---|
75 | .PP |
---|
76 | At the end of each item, consult will prompt for |
---|
77 | retrial of the same item, a new item, or end of session. |
---|
78 | When retrying an item, the previous values of queried |
---|
79 | attributes are displayed in square brackets. If the |
---|
80 | value is unchanged, reply with a carriage return; |
---|
81 | otherwise, specify the changed value. |
---|
82 | .PP |
---|
83 | Options and their meanings are: |
---|
84 | .PP |
---|
85 | .TP 12 |
---|
86 | .BI \-f filestem\^ |
---|
87 | Specify the filename stem (default |
---|
88 | .B DF) |
---|
89 | .TP |
---|
90 | .B \-t |
---|
91 | Display the decision tree at the start of the consulting |
---|
92 | session. |
---|
93 | .SH FILES |
---|
94 | .PP |
---|
95 | .in 8 |
---|
96 | consult |
---|
97 | .br |
---|
98 | filestem.names |
---|
99 | .br |
---|
100 | filestem.tree |
---|
101 | .in 0 |
---|
102 | .PP |
---|
103 | .SH SEE ALSO |
---|
104 | .PP |
---|
105 | c4.5(1) |
---|
106 | .PP |
---|
107 | .SH BUGS |
---|