see5-installer 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +8 -0
- data/.rubocop.yml +11 -0
- data/CHANGELOG.md +5 -0
- data/Gemfile +10 -0
- data/README.md +29 -0
- data/Rakefile +12 -0
- data/ext/c5.0/Makefile +86 -0
- data/ext/c5.0/attwinnow.c +394 -0
- data/ext/c5.0/c50.c +330 -0
- data/ext/c5.0/classify.c +700 -0
- data/ext/c5.0/confmat.c +195 -0
- data/ext/c5.0/construct.c +853 -0
- data/ext/c5.0/contin.c +613 -0
- data/ext/c5.0/defns.i +788 -0
- data/ext/c5.0/discr.c +307 -0
- data/ext/c5.0/extern.i +170 -0
- data/ext/c5.0/formrules.c +720 -0
- data/ext/c5.0/formtree.c +1158 -0
- data/ext/c5.0/getdata.c +521 -0
- data/ext/c5.0/getnames.c +733 -0
- data/ext/c5.0/global.c +211 -0
- data/ext/c5.0/gpl.txt +674 -0
- data/ext/c5.0/implicitatt.c +1112 -0
- data/ext/c5.0/info.c +146 -0
- data/ext/c5.0/mcost.c +138 -0
- data/ext/c5.0/modelfiles.c +952 -0
- data/ext/c5.0/p-thresh.c +313 -0
- data/ext/c5.0/prune.c +1069 -0
- data/ext/c5.0/report.c +345 -0
- data/ext/c5.0/rules.c +579 -0
- data/ext/c5.0/ruletree.c +398 -0
- data/ext/c5.0/siftrules.c +1285 -0
- data/ext/c5.0/sort.c +156 -0
- data/ext/c5.0/subset.c +599 -0
- data/ext/c5.0/text.i +223 -0
- data/ext/c5.0/trees.c +740 -0
- data/ext/c5.0/update.c +129 -0
- data/ext/c5.0/utility.c +1146 -0
- data/ext/c5.0/xval +150 -0
- data/ext/c5.0/xval.c +402 -0
- data/ext/gritbot/Makefile +98 -0
- data/ext/gritbot/check.c +1110 -0
- data/ext/gritbot/cluster.c +342 -0
- data/ext/gritbot/common.c +1269 -0
- data/ext/gritbot/continatt.c +412 -0
- data/ext/gritbot/defns.i +623 -0
- data/ext/gritbot/discratt.c +459 -0
- data/ext/gritbot/extern.i +101 -0
- data/ext/gritbot/getdata.c +329 -0
- data/ext/gritbot/getnames.c +573 -0
- data/ext/gritbot/global.c +104 -0
- data/ext/gritbot/gpl.txt +674 -0
- data/ext/gritbot/gritbot.c +295 -0
- data/ext/gritbot/implicitatt.c +1108 -0
- data/ext/gritbot/inspect.c +794 -0
- data/ext/gritbot/modelfiles.c +687 -0
- data/ext/gritbot/outlier.c +415 -0
- data/ext/gritbot/sort.c +130 -0
- data/ext/gritbot/text.i +159 -0
- data/ext/gritbot/update.c +126 -0
- data/ext/gritbot/utility.c +1029 -0
- data/ext/see5-installer/extconf.rb +25 -0
- data/lib/see5/installer.rb +10 -0
- data/lib/see5/installer/version.rb +7 -0
- data/see5-installer.gemspec +30 -0
- metadata +115 -0
data/ext/c5.0/defns.i
ADDED
@@ -0,0 +1,788 @@
|
|
1
|
+
/*************************************************************************/
|
2
|
+
/* */
|
3
|
+
/* Copyright 2010 Rulequest Research Pty Ltd. */
|
4
|
+
/* Author: Ross Quinlan (quinlan@rulequest.com) [Rev Jan 2016] */
|
5
|
+
/* */
|
6
|
+
/* This file is part of C5.0 GPL Edition, a single-threaded version */
|
7
|
+
/* of C5.0 release 2.07. */
|
8
|
+
/* */
|
9
|
+
/* C5.0 GPL Edition is free software: you can redistribute it and/or */
|
10
|
+
/* modify it under the terms of the GNU General Public License as */
|
11
|
+
/* published by the Free Software Foundation, either version 3 of the */
|
12
|
+
/* License, or (at your option) any later version. */
|
13
|
+
/* */
|
14
|
+
/* C5.0 GPL Edition is distributed in the hope that it will be useful, */
|
15
|
+
/* but WITHOUT ANY WARRANTY; without even the implied warranty of */
|
16
|
+
/* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU */
|
17
|
+
/* General Public License for more details. */
|
18
|
+
/* */
|
19
|
+
/* You should have received a copy of the GNU General Public License */
|
20
|
+
/* (gpl.txt) along with C5.0 GPL Edition. If not, see */
|
21
|
+
/* */
|
22
|
+
/* <http://www.gnu.org/licenses/>. */
|
23
|
+
/* */
|
24
|
+
/*************************************************************************/
|
25
|
+
|
26
|
+
|
27
|
+
|
28
|
+
/*************************************************************************/
|
29
|
+
/* */
|
30
|
+
/* Definitions used in C5.0 */
|
31
|
+
/* ------------------------ */
|
32
|
+
/* */
|
33
|
+
/*************************************************************************/
|
34
|
+
|
35
|
+
|
36
|
+
#define RELEASE "2.07 GPL Edition"
|
37
|
+
|
38
|
+
/* Uncomment following line to enable
|
39
|
+
sample estimates for large datasets.
|
40
|
+
This can lead to some variablility,
|
41
|
+
especially when used with SMP */
|
42
|
+
//#define SAMPLE_ESTIMATES
|
43
|
+
|
44
|
+
#include <stdio.h>
|
45
|
+
#include <math.h>
|
46
|
+
#include <string.h>
|
47
|
+
#include <stdlib.h>
|
48
|
+
#include <time.h>
|
49
|
+
#include <ctype.h>
|
50
|
+
#include <limits.h>
|
51
|
+
#include <float.h>
|
52
|
+
|
53
|
+
#include "text.i"
|
54
|
+
|
55
|
+
|
56
|
+
|
57
|
+
/*************************************************************************/
|
58
|
+
/* */
|
59
|
+
/* Definitions dependent on cc options */
|
60
|
+
/* */
|
61
|
+
/*************************************************************************/
|
62
|
+
|
63
|
+
|
64
|
+
#ifdef VerbOpt
|
65
|
+
#define Goodbye(x) {Cleanup(); exit(x);}
|
66
|
+
#else
|
67
|
+
#define Goodbye(x) exit(x)
|
68
|
+
#endif
|
69
|
+
|
70
|
+
#ifdef VerbOpt
|
71
|
+
#include <assert.h>
|
72
|
+
#define Verbosity(d,s) if(VERBOSITY >= d) {s;}
|
73
|
+
#else
|
74
|
+
#define assert(x)
|
75
|
+
#define Verbosity(d,s)
|
76
|
+
#endif
|
77
|
+
|
78
|
+
|
79
|
+
/* Alternative random number generator */
|
80
|
+
|
81
|
+
#define AltRandom drand48()
|
82
|
+
#define AltSeed(x) srand48(x)
|
83
|
+
|
84
|
+
#define Free(x) {free(x); x=0;}
|
85
|
+
|
86
|
+
|
87
|
+
/*************************************************************************/
|
88
|
+
/* */
|
89
|
+
/* Constants, macros etc. */
|
90
|
+
/* */
|
91
|
+
/*************************************************************************/
|
92
|
+
|
93
|
+
#define THEORYFRAC 0.23 /* discount rate for estimated coding cost */
|
94
|
+
|
95
|
+
#define Nil 0 /* null pointer */
|
96
|
+
#define false 0
|
97
|
+
#define true 1
|
98
|
+
#define None -1
|
99
|
+
#define Epsilon 1E-4
|
100
|
+
#define MinLeaf 0.05 /* minimum weight for non-null leaf */
|
101
|
+
#define Width 80 /* approx max width of output */
|
102
|
+
|
103
|
+
#define EXCLUDE 1 /* special attribute status: do not use */
|
104
|
+
#define SKIP 2 /* do not use in classifiers */
|
105
|
+
#define DISCRETE 4 /* ditto: collect values as data read */
|
106
|
+
#define ORDERED 8 /* ditto: ordered discrete values */
|
107
|
+
#define DATEVAL 16 /* ditto: YYYY/MM/DD or YYYY-MM-DD */
|
108
|
+
#define STIMEVAL 32 /* ditto: HH:MM:SS */
|
109
|
+
#define TSTMPVAL 64 /* date time */
|
110
|
+
|
111
|
+
#define CMINFO 1 /* generate confusion matrix */
|
112
|
+
#define USAGEINFO 2 /* print usage information */
|
113
|
+
|
114
|
+
/* unknown and N/A values are represented by
|
115
|
+
unlikely floating-point numbers
|
116
|
+
(octal 01600000000 and 01) */
|
117
|
+
#define UNKNOWN 01600000000 /* 1.5777218104420236e-30 */
|
118
|
+
#define NA 01 /* 1.4012984643248171e-45 */
|
119
|
+
|
120
|
+
#define BrDiscr 1
|
121
|
+
#define BrThresh 2
|
122
|
+
#define BrSubset 3
|
123
|
+
|
124
|
+
#define Plural(n) ((n) != 1 ? "s" : "")
|
125
|
+
|
126
|
+
#define AllocZero(N,T) (T *) Pcalloc(N, sizeof(T))
|
127
|
+
#define Alloc(N,T) AllocZero(N,T) /* for safety */
|
128
|
+
#define Realloc(V,N,T) V = (T *) Prealloc(V, (N)*sizeof(T))
|
129
|
+
|
130
|
+
#define Max(a,b) ((a)>(b) ? (a) : (b))
|
131
|
+
#define Min(a,b) ((a)<(b) ? (a) : (b))
|
132
|
+
|
133
|
+
#define Log2 0.69314718055994530942
|
134
|
+
#define Log(x) ((x) <= 0 ? 0.0 : log((double)x) / Log2)
|
135
|
+
|
136
|
+
#define Bit(b) (1 << (b))
|
137
|
+
#define In(b,s) ((s[(b) >> 3]) & Bit((b) & 07))
|
138
|
+
#define ClearBits(n,s) memset(s,0,n)
|
139
|
+
#define CopyBits(n,f,t) memcpy(t,f,n)
|
140
|
+
#define SetBit(b,s) (s[(b) >> 3] |= Bit((b) & 07))
|
141
|
+
#define ResetBit(b,s) (s[(b) >> 3] ^= Bit((b) & 07))
|
142
|
+
|
143
|
+
#define ForEach(v,f,l) for(v=f ; v<=l ; ++v)
|
144
|
+
|
145
|
+
#define CountCases(f,l) (UnitWeights ? (l-(f)+1.0) : SumWeights(f,l))
|
146
|
+
|
147
|
+
#define StatBit(a,b) (SpecialStatus[a]&(b))
|
148
|
+
#define Exclude(a) StatBit(a,EXCLUDE)
|
149
|
+
#define Skip(a) StatBit(a,EXCLUDE|SKIP)
|
150
|
+
#define Discrete(a) (MaxAttVal[a] || StatBit(a,DISCRETE))
|
151
|
+
#define Continuous(a) (! MaxAttVal[a] && ! StatBit(a,DISCRETE))
|
152
|
+
#define Ordered(a) StatBit(a,ORDERED)
|
153
|
+
#define DateVal(a) StatBit(a,DATEVAL)
|
154
|
+
#define TimeVal(a) StatBit(a,STIMEVAL)
|
155
|
+
#define TStampVal(a) StatBit(a,TSTMPVAL)
|
156
|
+
|
157
|
+
#define FreeUnlessNil(p) if((p)!=Nil) Free(p)
|
158
|
+
|
159
|
+
#define CheckClose(f) if(f) {fclose(f); f=Nil;}
|
160
|
+
|
161
|
+
#define Int(x) ((int)(x+0.5))
|
162
|
+
|
163
|
+
#define Space(s) (s == ' ' || s == '\n' || s == '\r' || s == '\t')
|
164
|
+
#define SkipComment while ( ( c = InChar(f) ) != '\n' && c != EOF )
|
165
|
+
|
166
|
+
#define P1(x) (rint((x)*10) / 10)
|
167
|
+
|
168
|
+
#define No(f,l) ((l)-(f)+1)
|
169
|
+
|
170
|
+
#define EmptyNA(T) (T->Branch[1]->Cases < 0.01)
|
171
|
+
|
172
|
+
#define Before(n1,n2) (n1->Tested < n2->Tested ||\
|
173
|
+
n1->Tested == n2->Tested && n1->Cut < n2->Cut)
|
174
|
+
|
175
|
+
#define Swap(a,b) {DataRec xab;\
|
176
|
+
assert(a >= 0 && a <= MaxCase &&\
|
177
|
+
b >= 0 && b <= MaxCase);\
|
178
|
+
xab = Case[a]; Case[a] = Case[b]; Case[b] = xab;}
|
179
|
+
|
180
|
+
|
181
|
+
#define NOFILE 0
|
182
|
+
#define BADCLASSTHRESH 1
|
183
|
+
#define LEQCLASSTHRESH 2
|
184
|
+
#define BADATTNAME 3
|
185
|
+
#define EOFINATT 4
|
186
|
+
#define SINGLEATTVAL 5
|
187
|
+
#define BADATTVAL 6
|
188
|
+
#define BADNUMBER 7
|
189
|
+
#define BADCLASS 8
|
190
|
+
#define BADCOSTCLASS 9
|
191
|
+
#define BADCOST 10
|
192
|
+
#define NOMEM 11
|
193
|
+
#define TOOMANYVALS 12
|
194
|
+
#define BADDISCRETE 13
|
195
|
+
#define NOTARGET 14
|
196
|
+
#define BADCTARGET 15
|
197
|
+
#define BADDTARGET 16
|
198
|
+
#define LONGNAME 17
|
199
|
+
#define HITEOF 18
|
200
|
+
#define MISSNAME 19
|
201
|
+
#define BADDATE 20
|
202
|
+
#define BADTIME 21
|
203
|
+
#define BADTSTMP 22
|
204
|
+
#define DUPATTNAME 23
|
205
|
+
#define UNKNOWNATT 24
|
206
|
+
#define BADDEF1 25
|
207
|
+
#define BADDEF2 26
|
208
|
+
#define BADDEF3 27
|
209
|
+
#define BADDEF4 28
|
210
|
+
#define SAMEATT 29
|
211
|
+
#define MODELFILE 30
|
212
|
+
#define CWTATTERR 31
|
213
|
+
|
214
|
+
#define READDATA 1
|
215
|
+
#define WINNOWATTS 2
|
216
|
+
#define FORMTREE 3
|
217
|
+
#define SIMPLIFYTREE 4
|
218
|
+
#define FORMRULES 5
|
219
|
+
#define SIFTRULES 6
|
220
|
+
#define EVALTRAIN 7
|
221
|
+
#define READTEST 8
|
222
|
+
#define EVALTEST 9
|
223
|
+
#define CLEANUP 10
|
224
|
+
#define ALLOCTABLES 11
|
225
|
+
#define RESULTS 12
|
226
|
+
#define READXDATA 13
|
227
|
+
|
228
|
+
|
229
|
+
/*************************************************************************/
|
230
|
+
/* */
|
231
|
+
/* Type definitions */
|
232
|
+
/* */
|
233
|
+
/*************************************************************************/
|
234
|
+
|
235
|
+
|
236
|
+
typedef unsigned char Boolean, BranchType, *Set, Byte;
|
237
|
+
typedef char *String;
|
238
|
+
|
239
|
+
typedef int CaseNo; /* data case number */
|
240
|
+
typedef float CaseCount; /* count of (partial) cases */
|
241
|
+
|
242
|
+
typedef int ClassNo, /* class number, 1..MaxClass */
|
243
|
+
DiscrValue, /* discrete attribute value */
|
244
|
+
Attribute; /* attribute number, 1..MaxAtt */
|
245
|
+
|
246
|
+
#ifdef USEDOUBLE
|
247
|
+
typedef double ContValue; /* continuous attribute value */
|
248
|
+
#define PREC 14 /* precision */
|
249
|
+
#else
|
250
|
+
typedef float ContValue; /* continuous attribute value */
|
251
|
+
#define PREC 7 /* precision */
|
252
|
+
#endif
|
253
|
+
|
254
|
+
|
255
|
+
typedef union _def_val
|
256
|
+
{
|
257
|
+
String _s_val; /* att val for comparison */
|
258
|
+
ContValue _n_val; /* number for arith */
|
259
|
+
}
|
260
|
+
DefVal;
|
261
|
+
|
262
|
+
typedef struct _def_elt
|
263
|
+
{
|
264
|
+
short _op_code; /* type of element */
|
265
|
+
DefVal _operand; /* string or numeric value */
|
266
|
+
}
|
267
|
+
DefElt, *Definition;
|
268
|
+
|
269
|
+
typedef struct _elt_rec
|
270
|
+
{
|
271
|
+
int Fi, /* index of first char of element */
|
272
|
+
Li; /* last ditto */
|
273
|
+
char Type; /* 'B', 'S', or 'N' */
|
274
|
+
}
|
275
|
+
EltRec;
|
276
|
+
|
277
|
+
#define DefOp(DE) DE._op_code
|
278
|
+
#define DefSVal(DE) DE._operand._s_val
|
279
|
+
#define DefNVal(DE) DE._operand._n_val
|
280
|
+
|
281
|
+
#define OP_ATT 0 /* opcodes */
|
282
|
+
#define OP_NUM 1
|
283
|
+
#define OP_STR 2
|
284
|
+
#define OP_MISS 3
|
285
|
+
#define OP_AND 10
|
286
|
+
#define OP_OR 11
|
287
|
+
#define OP_EQ 20
|
288
|
+
#define OP_NE 21
|
289
|
+
#define OP_GT 22
|
290
|
+
#define OP_GE 23
|
291
|
+
#define OP_LT 24
|
292
|
+
#define OP_LE 25
|
293
|
+
#define OP_SEQ 26
|
294
|
+
#define OP_SNE 27
|
295
|
+
#define OP_PLUS 30
|
296
|
+
#define OP_MINUS 31
|
297
|
+
#define OP_UMINUS 32
|
298
|
+
#define OP_MULT 33
|
299
|
+
#define OP_DIV 34
|
300
|
+
#define OP_MOD 35
|
301
|
+
#define OP_POW 36
|
302
|
+
#define OP_SIN 40
|
303
|
+
#define OP_COS 41
|
304
|
+
#define OP_TAN 42
|
305
|
+
#define OP_LOG 43
|
306
|
+
#define OP_EXP 44
|
307
|
+
#define OP_INT 45
|
308
|
+
#define OP_END 99
|
309
|
+
|
310
|
+
|
311
|
+
typedef union _attribute_value
|
312
|
+
{
|
313
|
+
DiscrValue _discr_val;
|
314
|
+
ContValue _cont_val;
|
315
|
+
}
|
316
|
+
AttValue, *DataRec;
|
317
|
+
|
318
|
+
typedef struct _sort_rec
|
319
|
+
{
|
320
|
+
ContValue V;
|
321
|
+
ClassNo C;
|
322
|
+
float W;
|
323
|
+
}
|
324
|
+
SortRec;
|
325
|
+
|
326
|
+
#define CVal(Case,Att) Case[Att]._cont_val
|
327
|
+
#define DVal(Case,Att) Case[Att]._discr_val
|
328
|
+
#define XDVal(Case,Att) (Case[Att]._discr_val & 077777777)
|
329
|
+
#define SVal(Case,Att) Case[Att]._discr_val
|
330
|
+
#define Class(Case) (*Case)._discr_val
|
331
|
+
#define Weight(Case) (*(Case-1))._cont_val
|
332
|
+
|
333
|
+
#define Unknown(Case,Att) (DVal(Case,Att)==UNKNOWN)
|
334
|
+
#define UnknownVal(AV) (AV._discr_val==UNKNOWN)
|
335
|
+
#define NotApplic(Case,Att) (Att != ClassAtt && DVal(Case,Att)==NA)
|
336
|
+
#define NotApplicVal(AV) (AV._discr_val==NA)
|
337
|
+
|
338
|
+
#define RelCWt(Case) ( Unknown(Case,CWtAtt)||\
|
339
|
+
NotApplic(Case,CWtAtt)||\
|
340
|
+
CVal(Case,CWtAtt)<=0 ? 1 :\
|
341
|
+
CVal(Case,CWtAtt)/AvCWt )
|
342
|
+
|
343
|
+
typedef struct _treerec *Tree;
|
344
|
+
typedef struct _treerec
|
345
|
+
{
|
346
|
+
BranchType NodeType;
|
347
|
+
ClassNo Leaf; /* best class at this node */
|
348
|
+
CaseCount Cases, /* no of cases at this node */
|
349
|
+
*ClassDist, /* class distribution of cases */
|
350
|
+
Errors; /* est or resub errors at this node */
|
351
|
+
Attribute Tested; /* attribute referenced in test */
|
352
|
+
int Forks, /* number of branches at this node */
|
353
|
+
Leaves; /* number of non-empty leaves in tree */
|
354
|
+
ContValue Cut, /* threshold for continuous attribute */
|
355
|
+
Lower, /* lower limit of soft threshold */
|
356
|
+
Upper, /* upper limit ditto */
|
357
|
+
Mid; /* midpoint for soft threshold */
|
358
|
+
Set *Subset; /* subsets of discrete values */
|
359
|
+
Tree *Branch, /* Branch[x] = subtree for outcome x */
|
360
|
+
Parent; /* node above this one */
|
361
|
+
}
|
362
|
+
TreeRec;
|
363
|
+
|
364
|
+
|
365
|
+
typedef struct _environment
|
366
|
+
{
|
367
|
+
CaseNo Xp, Ep; /* start and end of scan */
|
368
|
+
double Cases, /* total cases */
|
369
|
+
KnownCases, /* ditto less missing values */
|
370
|
+
ApplicCases, /* cases with numeric values */
|
371
|
+
HighCases, LowCases, /* cases above/below cut */
|
372
|
+
NAInfo, /* info for N/A values */
|
373
|
+
FixedSplitInfo, /* split info for ?, N/A */
|
374
|
+
BaseInfo, /* info before split */
|
375
|
+
UnknownRate, /* proportion of ? values */
|
376
|
+
MinSplit, /* min cases before/after cut */
|
377
|
+
**Freq, /* local Freq[4][class] */
|
378
|
+
*ClassFreq, /* local class frequencies */
|
379
|
+
*ValFreq; /* cases with val i */
|
380
|
+
ClassNo HighClass, LowClass; /* class after/before cut */
|
381
|
+
ContValue HighVal, LowVal; /* values after/before cut */
|
382
|
+
SortRec *SRec; /* for Cachesort() */
|
383
|
+
Set **Subset, /* Subset[att][number] */
|
384
|
+
*WSubset; /* working subsets */
|
385
|
+
int *Subsets, /* no of subsets for att */
|
386
|
+
Blocks, /* intermediate no of subsets */
|
387
|
+
Bytes, /* size of each subset */
|
388
|
+
ReasonableSubsets;
|
389
|
+
double *SubsetInfo, /* subset info */
|
390
|
+
*SubsetEntr, /* subset entropy */
|
391
|
+
**MergeInfo, /* info of merged subsets i,j */
|
392
|
+
**MergeEntr; /* entropy ditto */
|
393
|
+
}
|
394
|
+
EnvRec;
|
395
|
+
|
396
|
+
|
397
|
+
typedef int RuleNo; /* rule number */
|
398
|
+
|
399
|
+
typedef struct _condrec
|
400
|
+
{
|
401
|
+
BranchType NodeType; /* test type (see tree nodes) */
|
402
|
+
Attribute Tested; /* attribute tested */
|
403
|
+
ContValue Cut; /* threshold (if relevant) */
|
404
|
+
Set Subset; /* subset (if relevant) */
|
405
|
+
int TestValue, /* specified outcome of test */
|
406
|
+
TestI; /* rule tree index of this test */
|
407
|
+
}
|
408
|
+
CondRec, *Condition;
|
409
|
+
|
410
|
+
|
411
|
+
typedef struct _rulerec
|
412
|
+
{
|
413
|
+
RuleNo RNo; /* rule number */
|
414
|
+
int TNo, /* trial number */
|
415
|
+
Size; /* number of conditions */
|
416
|
+
Condition *Lhs; /* conditions themselves */
|
417
|
+
ClassNo Rhs; /* class given by rule */
|
418
|
+
CaseCount Cover, /* number of cases covered by rule */
|
419
|
+
Correct; /* number on which correct */
|
420
|
+
float Prior; /* prior probability of RHS */
|
421
|
+
int Vote; /* unit = 0.001 */
|
422
|
+
}
|
423
|
+
RuleRec, *CRule;
|
424
|
+
|
425
|
+
|
426
|
+
typedef struct _ruletreerec *RuleTree;
|
427
|
+
typedef struct _ruletreerec
|
428
|
+
{
|
429
|
+
RuleNo *Fire; /* rules matched at this node */
|
430
|
+
Condition CondTest; /* new test */
|
431
|
+
int Forks; /* number of branches */
|
432
|
+
RuleTree *Branch; /* subtrees */
|
433
|
+
}
|
434
|
+
RuleTreeRec;
|
435
|
+
|
436
|
+
|
437
|
+
typedef struct _rulesetrec
|
438
|
+
{
|
439
|
+
RuleNo SNRules; /* number of rules */
|
440
|
+
CRule *SRule; /* rules */
|
441
|
+
ClassNo SDefault; /* default class for this ruleset */
|
442
|
+
RuleTree RT; /* rule tree (see ruletree.c) */
|
443
|
+
}
|
444
|
+
RuleSetRec, *CRuleSet;
|
445
|
+
|
446
|
+
|
447
|
+
|
448
|
+
/*************************************************************************/
|
449
|
+
/* */
|
450
|
+
/* Function prototypes */
|
451
|
+
/* */
|
452
|
+
/*************************************************************************/
|
453
|
+
|
454
|
+
/* c50.c */
|
455
|
+
|
456
|
+
int main(int, char *[]);
|
457
|
+
void FreeClassifier(int Trial);
|
458
|
+
|
459
|
+
/* construct.c */
|
460
|
+
|
461
|
+
void ConstructClassifiers(void);
|
462
|
+
void InitialiseWeights(void);
|
463
|
+
void SetAvCWt(void);
|
464
|
+
void Evaluate(int Flags);
|
465
|
+
void EvaluateSingle(int Flags);
|
466
|
+
void EvaluateBoost(int Flags);
|
467
|
+
void RecordAttUsage(DataRec Case, int *Usage);
|
468
|
+
|
469
|
+
/* getnames.c */
|
470
|
+
|
471
|
+
Boolean ReadName(FILE *f, String s, int n, char ColonOpt);
|
472
|
+
void GetNames(FILE *Nf);
|
473
|
+
void ExplicitAtt(FILE *Nf);
|
474
|
+
int Which(String Val, String *List, int First, int Last);
|
475
|
+
void ListAttsUsed(void);
|
476
|
+
void FreeNames(void);
|
477
|
+
int InChar(FILE *f);
|
478
|
+
|
479
|
+
/* implicitatt.c */
|
480
|
+
|
481
|
+
void ImplicitAtt(FILE *Nf);
|
482
|
+
void ReadDefinition(FILE *f);
|
483
|
+
void Append(char c);
|
484
|
+
Boolean Expression(void);
|
485
|
+
Boolean Conjunct(void);
|
486
|
+
Boolean SExpression(void);
|
487
|
+
Boolean AExpression(void);
|
488
|
+
Boolean Term(void);
|
489
|
+
Boolean Factor(void);
|
490
|
+
Boolean Primary(void);
|
491
|
+
Boolean Atom(void);
|
492
|
+
Boolean Find(String S);
|
493
|
+
int FindOne(String *Alt);
|
494
|
+
Attribute FindAttName(void);
|
495
|
+
void DefSyntaxError(String Msg);
|
496
|
+
void DefSemanticsError(int Fi, String Msg, int OpCode);
|
497
|
+
void Dump(char OpCode, ContValue F, String S, int Fi);
|
498
|
+
void DumpOp(char OpCode, int Fi);
|
499
|
+
Boolean UpdateTStack(char OpCode, ContValue F, String S, int Fi);
|
500
|
+
AttValue EvaluateDef(Definition D, DataRec Case);
|
501
|
+
|
502
|
+
/* getdata.c */
|
503
|
+
|
504
|
+
void GetData(FILE *Df, Boolean Train, Boolean AllowUnknownClass);
|
505
|
+
DataRec GetDataRec(FILE *Df, Boolean Train);
|
506
|
+
CaseNo CountData(FILE *Df);
|
507
|
+
int StoreIVal(String s);
|
508
|
+
void FreeData(void);
|
509
|
+
void CheckValue(DataRec Case, Attribute Att);
|
510
|
+
|
511
|
+
/* mcost.c */
|
512
|
+
|
513
|
+
void GetMCosts(FILE *f);
|
514
|
+
|
515
|
+
/* attwinnow.c */
|
516
|
+
|
517
|
+
void WinnowAtts(void);
|
518
|
+
float TrialTreeCost(Boolean FirstTime);
|
519
|
+
float ErrCost(Tree T, CaseNo Fp, CaseNo Lp);
|
520
|
+
void ScanTree(Tree T, Boolean *Used);
|
521
|
+
|
522
|
+
/* formtree.c */
|
523
|
+
|
524
|
+
void InitialiseTreeData(void);
|
525
|
+
void FreeTreeData(void);
|
526
|
+
void SetMinGainThresh(void);
|
527
|
+
void FormTree(CaseNo, CaseNo, int, Tree *);
|
528
|
+
void SampleEstimate(CaseNo Fp, CaseNo Lp, CaseCount Cases);
|
529
|
+
void Sample(CaseNo Fp, CaseNo Lp, CaseNo N);
|
530
|
+
Attribute ChooseSplit(CaseNo Fp, CaseNo Lp, CaseCount Cases, Boolean Sampled);
|
531
|
+
void ProcessQueue(CaseNo WFp, CaseNo WLp, CaseCount WCases);
|
532
|
+
Attribute FindBestAtt(CaseCount Cases);
|
533
|
+
void EvalDiscrSplit(Attribute Att, CaseCount Cases);
|
534
|
+
CaseNo Group(DiscrValue, CaseNo, CaseNo, Tree);
|
535
|
+
CaseCount SumWeights(CaseNo, CaseNo);
|
536
|
+
CaseCount SumNocostWeights(CaseNo, CaseNo);
|
537
|
+
void FindClassFreq(double [], CaseNo, CaseNo);
|
538
|
+
void FindAllFreq(CaseNo, CaseNo);
|
539
|
+
void Divide(Tree Node, CaseNo Fp, CaseNo Lp, int Level);
|
540
|
+
|
541
|
+
/* discr.c */
|
542
|
+
|
543
|
+
void EvalDiscreteAtt(Attribute Att, CaseCount Cases);
|
544
|
+
void EvalOrderedAtt(Attribute Att, CaseCount Cases);
|
545
|
+
void SetDiscrFreq(Attribute Att);
|
546
|
+
double DiscrKnownBaseInfo(CaseCount KnownCases, DiscrValue MaxVal);
|
547
|
+
void DiscreteTest(Tree Node, Attribute Att);
|
548
|
+
|
549
|
+
/* contin.c */
|
550
|
+
|
551
|
+
void EvalContinuousAtt(Attribute Att, CaseNo Fp, CaseNo Lp);
|
552
|
+
void EstimateMaxGR(Attribute Att, CaseNo Fp, CaseNo Lp);
|
553
|
+
void PrepareForContin(Attribute Att, CaseNo Fp, CaseNo Lp);
|
554
|
+
CaseNo PrepareForScan(CaseNo Lp);
|
555
|
+
void ContinTest(Tree Node, Attribute Att);
|
556
|
+
void AdjustAllThresholds(Tree T);
|
557
|
+
void AdjustThresholds(Tree T, Attribute Att, CaseNo *Ep);
|
558
|
+
ContValue GreatestValueBelow(ContValue Th, CaseNo *Ep);
|
559
|
+
|
560
|
+
/* info.c */
|
561
|
+
|
562
|
+
double ComputeGain(double BaseInfo, float UnknFrac, DiscrValue MaxVal,
|
563
|
+
CaseCount TotalCases);
|
564
|
+
double TotalInfo(double V[], DiscrValue MinVal, DiscrValue MaxVal);
|
565
|
+
void PrintDistribution(Attribute Att, DiscrValue MinVal,
|
566
|
+
DiscrValue MaxVal, double **Freq, double *ValFreq,
|
567
|
+
Boolean ShowNames);
|
568
|
+
|
569
|
+
/* subset.c */
|
570
|
+
|
571
|
+
void InitialiseBellNumbers(void);
|
572
|
+
void EvalSubset(Attribute Att, CaseCount Cases);
|
573
|
+
void Merge(DiscrValue x, DiscrValue y, CaseCount Cases);
|
574
|
+
void EvaluatePair(DiscrValue x, DiscrValue y, CaseCount Cases);
|
575
|
+
void PrintSubset(Attribute Att, Set Ss);
|
576
|
+
void SubsetTest(Tree Node, Attribute Att);
|
577
|
+
Boolean SameDistribution(DiscrValue V1, DiscrValue V2);
|
578
|
+
void AddBlock(DiscrValue V1, DiscrValue V2);
|
579
|
+
void AddBlock(DiscrValue V1, DiscrValue V2);
|
580
|
+
void MoveBlock(DiscrValue V1, DiscrValue V2);
|
581
|
+
|
582
|
+
/* prune.c */
|
583
|
+
|
584
|
+
void Prune(Tree T);
|
585
|
+
void EstimateErrs(Tree T, CaseNo Fp, CaseNo Lp, int Sh, int Flags);
|
586
|
+
void GlobalPrune(Tree T);
|
587
|
+
void FindMinCC(Tree T);
|
588
|
+
void InsertParents(Tree T, Tree P);
|
589
|
+
void CheckSubsets(Tree T, Boolean);
|
590
|
+
void InitialiseExtraErrs(void);
|
591
|
+
float ExtraErrs(CaseCount N, CaseCount E, ClassNo C);
|
592
|
+
float RawExtraErrs(CaseCount N, CaseCount E);
|
593
|
+
void RestoreDistribs(Tree T);
|
594
|
+
void CompressBranches(Tree T);
|
595
|
+
void SetGlobalUnitWeights(int LocalFlag);
|
596
|
+
|
597
|
+
/* p-thresh.c */
|
598
|
+
|
599
|
+
void SoftenThresh(Tree T);
|
600
|
+
void ResubErrs(Tree T, CaseNo Fp, CaseNo Lp);
|
601
|
+
void FindBounds(Tree T, CaseNo Fp, CaseNo Lp);
|
602
|
+
|
603
|
+
/* classify.c */
|
604
|
+
|
605
|
+
ClassNo TreeClassify(DataRec Case, Tree DecisionTree);
|
606
|
+
void FollowAllBranches(DataRec Case, Tree T, float Fraction);
|
607
|
+
ClassNo RuleClassify(DataRec Case, CRuleSet RS);
|
608
|
+
int FindOutcome(DataRec Case, Condition OneCond);
|
609
|
+
Boolean Matches(CRule R, DataRec Case);
|
610
|
+
void CheckActiveSpace(int N);
|
611
|
+
void MarkActive(RuleTree RT, DataRec Case);
|
612
|
+
void SortActive(void);
|
613
|
+
void CheckUtilityBand(int *u, RuleNo r, ClassNo Actual, ClassNo Default);
|
614
|
+
ClassNo BoostClassify(DataRec Case, int MaxTrial);
|
615
|
+
ClassNo SelectClass(ClassNo Default, Boolean UseCosts);
|
616
|
+
ClassNo Classify(DataRec Case);
|
617
|
+
float Interpolate(Tree T, ContValue Val);
|
618
|
+
|
619
|
+
/* special case for dual-purpose routines */
|
620
|
+
|
621
|
+
void FindLeaf(DataRec Case, Tree T, Tree PT, float Wt);
|
622
|
+
Boolean Satisfies(DataRec Case, Condition OneCond);
|
623
|
+
|
624
|
+
/* sort.c */
|
625
|
+
|
626
|
+
void Quicksort(CaseNo Fp, CaseNo Lp, Attribute Att);
|
627
|
+
void Cachesort(CaseNo Fp, CaseNo Lp, SortRec *SRec);
|
628
|
+
|
629
|
+
/* trees.c */
|
630
|
+
|
631
|
+
void FindDepth(Tree T);
|
632
|
+
void PrintTree(Tree T, String Title);
|
633
|
+
void Show(Tree T, int Sh);
|
634
|
+
void ShowBranch(int Sh, Tree T, DiscrValue v, DiscrValue BrNo);
|
635
|
+
DiscrValue Elements(Attribute Att, Set S, DiscrValue *Last);
|
636
|
+
int MaxLine(Tree SubTree);
|
637
|
+
void Indent(int Sh, int BrNo);
|
638
|
+
void FreeTree(Tree T);
|
639
|
+
Tree Leaf(double *Freq, ClassNo NodeClass, CaseCount Cases,
|
640
|
+
CaseCount Errors);
|
641
|
+
void Sprout(Tree T, DiscrValue Branches);
|
642
|
+
void UnSprout(Tree T);
|
643
|
+
int TreeSize(Tree T);
|
644
|
+
int ExpandedLeafCount(Tree T);
|
645
|
+
int TreeDepth(Tree T);
|
646
|
+
Tree CopyTree(Tree T);
|
647
|
+
|
648
|
+
/* utility.c */
|
649
|
+
|
650
|
+
void PrintHeader(String Title);
|
651
|
+
char ProcessOption(int Argc, char **Argv, char *Str);
|
652
|
+
void *Pmalloc(size_t Bytes);
|
653
|
+
void *Prealloc(void *Present, size_t Bytes);
|
654
|
+
void *Pcalloc(size_t Number, unsigned int Size);
|
655
|
+
void FreeVector(void **V, int First, int Last);
|
656
|
+
DataRec NewCase(void);
|
657
|
+
void FreeCases(void);
|
658
|
+
void FreeLastCase(DataRec Case);
|
659
|
+
double KRandom(void);
|
660
|
+
void ResetKR(int KRInit);
|
661
|
+
void Error(int ErrNo, String S1, String S2);
|
662
|
+
String CaseLabel(CaseNo N);
|
663
|
+
FILE * GetFile(String Extension, String RW);
|
664
|
+
double ExecTime(void);
|
665
|
+
int Denominator(ContValue Val);
|
666
|
+
int GetInt(String S, int N);
|
667
|
+
int DateToDay(String DS);
|
668
|
+
void DayToDate(int DI, String Date);
|
669
|
+
int TimeToSecs(String TS);
|
670
|
+
void SecsToTime(int Secs, String Time);
|
671
|
+
void SetTSBase(int y);
|
672
|
+
int TStampToMins(String TS);
|
673
|
+
void Check(float Val, float Low, float High);
|
674
|
+
void CValToStr(ContValue CV, Attribute Att, String DS);
|
675
|
+
double rint(double v);
|
676
|
+
void Cleanup(void);
|
677
|
+
#ifdef UTF8
|
678
|
+
int UTF8CharWidth(unsigned char *U);
|
679
|
+
int wcwidth(wchar_t ucs);
|
680
|
+
int wcswidth(const wchar_t *pwcs, size_t n);
|
681
|
+
#endif
|
682
|
+
|
683
|
+
/* confmat.c */
|
684
|
+
|
685
|
+
void PrintConfusionMatrix(CaseNo *ConfusionMat);
|
686
|
+
void PrintErrorBreakdown(CaseNo *ConfusionMat);
|
687
|
+
void PrintUsageInfo(CaseNo *Usage);
|
688
|
+
|
689
|
+
/* formrules.c */
|
690
|
+
|
691
|
+
CRuleSet FormRules(Tree T);
|
692
|
+
void Scan(Tree T);
|
693
|
+
void SetupNCost(void);
|
694
|
+
void PushCondition(void);
|
695
|
+
void PopCondition(void);
|
696
|
+
void PruneRule(Condition Cond[], ClassNo TargetClass);
|
697
|
+
void ProcessLists(void);
|
698
|
+
void AddToList(CaseNo *List, CaseNo N);
|
699
|
+
void DeleteFromList(CaseNo *Before, CaseNo N);
|
700
|
+
int SingleFail(CaseNo i);
|
701
|
+
void Increment(int d, CaseNo i, double *Total, double *Errors);
|
702
|
+
void FreeFormRuleData(void);
|
703
|
+
|
704
|
+
/* rules.c */
|
705
|
+
|
706
|
+
Boolean NewRule(Condition Cond[], int NConds, ClassNo TargetClass,
|
707
|
+
Boolean *Deleted, CRule Existing,
|
708
|
+
CaseCount Cover, CaseCount Correct, float Prior);
|
709
|
+
void ListSort(int *L, int Fp, int Lp);
|
710
|
+
Byte *Compress(int *L);
|
711
|
+
void Uncompress(Byte *CL, int *UCL);
|
712
|
+
Boolean SameRule(RuleNo r, Condition Cond[], int NConds,
|
713
|
+
ClassNo TargetClass);
|
714
|
+
void FreeRule(CRule R);
|
715
|
+
void FreeRules(CRuleSet RS);
|
716
|
+
void PrintRules(CRuleSet, String);
|
717
|
+
void PrintRule(CRule R);
|
718
|
+
void PrintCondition(Condition C);
|
719
|
+
|
720
|
+
/* siftrules.c */
|
721
|
+
|
722
|
+
void SiftRules(float EstErrRate);
|
723
|
+
void InvertFires(void);
|
724
|
+
void FindTestCodes(void);
|
725
|
+
float CondBits(Condition C);
|
726
|
+
void SetInitialTheory(void);
|
727
|
+
void CoverClass(ClassNo Target);
|
728
|
+
double MessageLength(RuleNo NR, double RuleBits, float Errs);
|
729
|
+
void HillClimb(void);
|
730
|
+
void InitialiseVotes(void);
|
731
|
+
void CountVotes(CaseNo i);
|
732
|
+
void UpdateDeltaErrs(CaseNo i, double Delta, RuleNo Toggle);
|
733
|
+
CaseCount CalculateDeltaErrs(void);
|
734
|
+
void PruneSubsets(void);
|
735
|
+
void SetDefaultClass(void);
|
736
|
+
void SwapRule(RuleNo A, RuleNo B);
|
737
|
+
int OrderByUtility(void);
|
738
|
+
int OrderByClass(void);
|
739
|
+
void OrderRules(void);
|
740
|
+
void GenerateLogs(int MaxN);
|
741
|
+
void FreeSiftRuleData(void);
|
742
|
+
|
743
|
+
/* ruletree.c */
|
744
|
+
|
745
|
+
void ConstructRuleTree(CRuleSet RS);
|
746
|
+
void SetTestIndex(Condition C);
|
747
|
+
RuleTree GrowRT(RuleNo *RR, int RRN, CRule *Rule);
|
748
|
+
int DesiredOutcome(CRule R, int TI);
|
749
|
+
int SelectTest(RuleNo *RR, int RRN, CRule *Rule);
|
750
|
+
void FreeRuleTree(RuleTree RT);
|
751
|
+
|
752
|
+
/* modelfiles.c */
|
753
|
+
|
754
|
+
void CheckFile(String Extension, Boolean Write);
|
755
|
+
void WriteFilePrefix(String Extension);
|
756
|
+
void ReadFilePrefix(String Extension);
|
757
|
+
void SaveDiscreteNames(void);
|
758
|
+
void SaveTree(Tree T, String Extension);
|
759
|
+
void OutTree(Tree T);
|
760
|
+
void SaveRules(CRuleSet RS, String Extension);
|
761
|
+
void AsciiOut(String Pre, String S);
|
762
|
+
void ReadHeader(void);
|
763
|
+
Tree GetTree(String Extension);
|
764
|
+
Tree InTree(void);
|
765
|
+
CRuleSet GetRules(String Extension);
|
766
|
+
CRuleSet InRules(void);
|
767
|
+
CRule InRule(void);
|
768
|
+
Condition InCondition(void);
|
769
|
+
int ReadProp(char *Delim);
|
770
|
+
String RemoveQuotes(String S);
|
771
|
+
Set MakeSubset(Attribute Att);
|
772
|
+
void StreamIn(String S, int n);
|
773
|
+
|
774
|
+
/* update.c (Unix) or winmain.c (WIN32) */
|
775
|
+
|
776
|
+
void NotifyStage(int);
|
777
|
+
void Progress(float);
|
778
|
+
|
779
|
+
/* xval.c */
|
780
|
+
|
781
|
+
void CrossVal(void);
|
782
|
+
void Prepare(void);
|
783
|
+
void Shuffle(int *Vec);
|
784
|
+
void Summary(void);
|
785
|
+
float SE(float sum, float sumsq, int no);
|
786
|
+
|
787
|
+
|
788
|
+
|