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
@@ -0,0 +1,794 @@
|
|
1
|
+
/*************************************************************************/
|
2
|
+
/* */
|
3
|
+
/* Copyright 2010 Rulequest Research Pty Ltd. */
|
4
|
+
/* */
|
5
|
+
/* This file is part of GritBot GPL Edition, a single-threaded version */
|
6
|
+
/* of GritBot release 2.01. */
|
7
|
+
/* */
|
8
|
+
/* GritBot GPL Edition is free software: you can redistribute it */
|
9
|
+
/* and/or modify it under the terms of the GNU General Public License */
|
10
|
+
/* as published by the Free Software Foundation, either version 3 of */
|
11
|
+
/* the License, or (at your option) any later version. */
|
12
|
+
/* */
|
13
|
+
/* GritBot GPL Edition is distributed in the hope that it will be */
|
14
|
+
/* useful, but WITHOUT ANY WARRANTY; without even the implied warranty */
|
15
|
+
/* of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the */
|
16
|
+
/* GNU General Public License for more details. */
|
17
|
+
/* */
|
18
|
+
/* You should have received a copy of the GNU General Public License */
|
19
|
+
/* (gpl.txt) along with GritBot GPL Edition. If not, see */
|
20
|
+
/* */
|
21
|
+
/* <http://www.gnu.org/licenses/>. */
|
22
|
+
/* */
|
23
|
+
/*************************************************************************/
|
24
|
+
|
25
|
+
|
26
|
+
/*************************************************************************/
|
27
|
+
/* */
|
28
|
+
/* Inspect: Use saved .sift file to check new cases. */
|
29
|
+
/* ------------------------------------------------- */
|
30
|
+
/* */
|
31
|
+
/*************************************************************************/
|
32
|
+
|
33
|
+
|
34
|
+
#include "defns.i"
|
35
|
+
|
36
|
+
|
37
|
+
/*************************************************************************/
|
38
|
+
/* */
|
39
|
+
/* Global Data */
|
40
|
+
/* ----------- */
|
41
|
+
/* */
|
42
|
+
/*************************************************************************/
|
43
|
+
|
44
|
+
Attribute ClassAtt=0, /* attribute to use as class */
|
45
|
+
LabelAtt; /* attribute to use as case ID */
|
46
|
+
|
47
|
+
int MaxAtt, /* max att number */
|
48
|
+
MaxDiscrVal=3, /* max discrete values for any att */
|
49
|
+
MaxLabel=0, /* max characters in case label */
|
50
|
+
LineNo=0, /* input line number */
|
51
|
+
ErrMsgs=0, /* errors found */
|
52
|
+
TSBase=0; /* base day for time stamps */
|
53
|
+
|
54
|
+
CaseNo MaxCase=-1, /* max data item number */
|
55
|
+
LastDataCase, /* max item in .data file */
|
56
|
+
*GLp; /* last case stack */
|
57
|
+
|
58
|
+
Description *Case=Nil, /* data items */
|
59
|
+
*SaveCase=Nil; /* items in original order */
|
60
|
+
|
61
|
+
DiscrValue *MaxAttVal=Nil; /* number of values for each att */
|
62
|
+
|
63
|
+
char *SpecialStatus=Nil; /* special att treatment */
|
64
|
+
|
65
|
+
Boolean *UseLogs=Nil, /* use log transformation */
|
66
|
+
*SomeNA=Nil; /* att has missing values */
|
67
|
+
|
68
|
+
Definition *AttDef=Nil; /* definitions of implicit atts */
|
69
|
+
|
70
|
+
String *AttName=Nil, /* att names */
|
71
|
+
**AttValName=Nil; /* att value names */
|
72
|
+
|
73
|
+
Boolean SIFT=true, /* write sift file */
|
74
|
+
LIST=false, /* list case numbers */
|
75
|
+
TargetSaved; /* has current classatt been saved? */
|
76
|
+
|
77
|
+
int MAXOUT=0; /* max anoms reported */
|
78
|
+
|
79
|
+
float MINABNORM=8; /* SDs for abnormal value */
|
80
|
+
|
81
|
+
unsigned char *Prec=Nil; /* [att] */
|
82
|
+
int LastLevel=0;
|
83
|
+
|
84
|
+
Set LeftSS=Nil; /* temporary subset */
|
85
|
+
|
86
|
+
float *Surprise=Nil; /* temporary DProb values */
|
87
|
+
|
88
|
+
char Fn[512]; /* file name */
|
89
|
+
|
90
|
+
ContValue *LowTail=Nil, /* lowest value analysed */
|
91
|
+
*HighTail=Nil; /* highest ditto */
|
92
|
+
|
93
|
+
Clust *Cluster=Nil; /* clusters found */
|
94
|
+
int NClust=0,
|
95
|
+
ClustSpace=0;
|
96
|
+
|
97
|
+
CaveatRec *Caveat=Nil; /* caveat records */
|
98
|
+
int NCaveat; /* active caveats */
|
99
|
+
|
100
|
+
EnvRec GEnv; /* sift environment */
|
101
|
+
|
102
|
+
FILE *Sf=0; /* sift file */
|
103
|
+
|
104
|
+
String FileStem="undefined";
|
105
|
+
|
106
|
+
|
107
|
+
#define SetIOpt(V) V = strtol(OptArg, &EndPtr, 10);\
|
108
|
+
if ( ! EndPtr || *EndPtr != '\00' ) break;\
|
109
|
+
ArgOK = true
|
110
|
+
|
111
|
+
|
112
|
+
|
113
|
+
/*************************************************************************/
|
114
|
+
/* */
|
115
|
+
/* Main */
|
116
|
+
/* */
|
117
|
+
/*************************************************************************/
|
118
|
+
|
119
|
+
|
120
|
+
int main(int Argc, char *Argv[])
|
121
|
+
/* ---- */
|
122
|
+
{
|
123
|
+
CaseNo i;
|
124
|
+
FILE *F;
|
125
|
+
int o;
|
126
|
+
char *EndPtr;
|
127
|
+
Boolean FirstTime=true, ArgOK;
|
128
|
+
double StartTime;
|
129
|
+
extern String OptArg, Option;
|
130
|
+
|
131
|
+
SIFT = false; /* important for CleanUp()! */
|
132
|
+
|
133
|
+
StartTime = ExecTime();
|
134
|
+
PrintHeader(" Inspector");
|
135
|
+
|
136
|
+
/* Process options */
|
137
|
+
|
138
|
+
while ( (o = ProcessOption(Argc, Argv, "f+n+rh")) )
|
139
|
+
{
|
140
|
+
if ( FirstTime )
|
141
|
+
{
|
142
|
+
fprintf(Of, "\n " T_Options ":\n");
|
143
|
+
FirstTime = false;
|
144
|
+
}
|
145
|
+
|
146
|
+
ArgOK = false;
|
147
|
+
|
148
|
+
switch (o)
|
149
|
+
{
|
150
|
+
case 'f': FileStem = OptArg;
|
151
|
+
fprintf(Of, F_Application, FileStem);
|
152
|
+
ArgOK = true;
|
153
|
+
break;
|
154
|
+
|
155
|
+
case 'n': SetIOpt(MAXOUT);
|
156
|
+
fprintf(Of, F_MaxOut, MAXOUT);
|
157
|
+
Check(MAXOUT, 1, 1000000);
|
158
|
+
break;
|
159
|
+
|
160
|
+
case 'r': LIST = true;
|
161
|
+
fprintf(Of, F_ListAnoms);
|
162
|
+
ArgOK = true;
|
163
|
+
break;
|
164
|
+
|
165
|
+
case '?': printf(" **Unrecognised option %s\n", Option);
|
166
|
+
exit(1);
|
167
|
+
}
|
168
|
+
|
169
|
+
if ( ! ArgOK )
|
170
|
+
{
|
171
|
+
if ( o != 'h' )
|
172
|
+
{
|
173
|
+
fprintf(Of, F_UnrecogOpt, Option);
|
174
|
+
}
|
175
|
+
fprintf(Of, F_CkOptList);
|
176
|
+
exit(1);
|
177
|
+
}
|
178
|
+
}
|
179
|
+
|
180
|
+
/* Open the .sift file and recover attribute information */
|
181
|
+
|
182
|
+
CheckFile(".sift", false);
|
183
|
+
fprintf(Of, F_ReadSift, FileStem);
|
184
|
+
|
185
|
+
if ( ! (F = GetFile(".cases", "r")) ) Error(NOFILE, "", "");
|
186
|
+
GetData(F, true);
|
187
|
+
fprintf(Of, F_ReadCases(MaxCase+1, MaxAtt, FileStem));
|
188
|
+
LastDataCase = MaxCase;
|
189
|
+
|
190
|
+
/* Remember original case order */
|
191
|
+
|
192
|
+
SaveCase = Alloc(MaxCase+1, Description);
|
193
|
+
ForEach(i, 0, MaxCase)
|
194
|
+
{
|
195
|
+
SaveCase[i] = Case[i];
|
196
|
+
}
|
197
|
+
|
198
|
+
/* Check the cases for anomalies using information recorded
|
199
|
+
in the .sift file */
|
200
|
+
|
201
|
+
ProcessSift();
|
202
|
+
|
203
|
+
/* Restore original case order before reporting potential anomalies */
|
204
|
+
|
205
|
+
Free(Case);
|
206
|
+
Case = SaveCase;
|
207
|
+
SaveCase = Nil;
|
208
|
+
|
209
|
+
ReportOutliers();
|
210
|
+
|
211
|
+
fprintf(Of, F_Time(ExecTime() - StartTime));
|
212
|
+
|
213
|
+
CleanupSift();
|
214
|
+
|
215
|
+
return 0;
|
216
|
+
}
|
217
|
+
|
218
|
+
|
219
|
+
|
220
|
+
/*************************************************************************/
|
221
|
+
/* */
|
222
|
+
/* Deallocate all dynamic storage */
|
223
|
+
/* */
|
224
|
+
/*************************************************************************/
|
225
|
+
|
226
|
+
|
227
|
+
void CleanupSift()
|
228
|
+
/* ----------- */
|
229
|
+
{
|
230
|
+
Attribute Att;
|
231
|
+
int i, d;
|
232
|
+
|
233
|
+
if ( Sf )
|
234
|
+
{
|
235
|
+
fclose(Sf); Sf=0;
|
236
|
+
}
|
237
|
+
|
238
|
+
/* Any stuff from ProcessSift() */
|
239
|
+
|
240
|
+
FreeUnlessNil(LeftSS); LeftSS=Nil;
|
241
|
+
FreeUnlessNil(Surprise); Surprise=Nil;
|
242
|
+
FreeUnlessNil(SomeNA); SomeNA=Nil;
|
243
|
+
FreeUnlessNil(GLp); GLp=Nil;
|
244
|
+
FreeUnlessNil(Prec); Prec=Nil;
|
245
|
+
FreeUnlessNil(UseLogs); UseLogs=Nil;
|
246
|
+
FreeUnlessNil(LowTail); LowTail=Nil;
|
247
|
+
FreeUnlessNil(HighTail); HighTail=Nil;
|
248
|
+
|
249
|
+
if ( Caveat )
|
250
|
+
{
|
251
|
+
ForEach(Att, 1, MaxAtt)
|
252
|
+
{
|
253
|
+
FreeUnlessNil(Caveat[Att-1].Subset);
|
254
|
+
}
|
255
|
+
Free(Caveat); Caveat=Nil;
|
256
|
+
}
|
257
|
+
|
258
|
+
Free(GEnv.Possible);
|
259
|
+
Free(GEnv.Tested);
|
260
|
+
ForEach(i, 0, GEnv.MaxLevel-1)
|
261
|
+
{
|
262
|
+
FreeUnlessNil(GEnv.Test[i].Left);
|
263
|
+
}
|
264
|
+
Free(GEnv.Test);
|
265
|
+
|
266
|
+
if ( MaxCase >= 0 )
|
267
|
+
{
|
268
|
+
FreeData(); MaxCase=-1;
|
269
|
+
}
|
270
|
+
FreeUnlessNil(SaveCase); SaveCase=Nil;
|
271
|
+
|
272
|
+
FreeNames();
|
273
|
+
|
274
|
+
if ( NClust > 0 )
|
275
|
+
{
|
276
|
+
ForEach(d, 0, NClust-1)
|
277
|
+
{
|
278
|
+
FreeClust(Cluster[d]);
|
279
|
+
}
|
280
|
+
}
|
281
|
+
NClust = ClustSpace = 0;
|
282
|
+
FreeUnlessNil(Cluster); Cluster=Nil;
|
283
|
+
|
284
|
+
NotifyStage(0);
|
285
|
+
Progress(0);
|
286
|
+
}
|
287
|
+
|
288
|
+
|
289
|
+
|
290
|
+
/*************************************************************************/
|
291
|
+
/* */
|
292
|
+
/* Check entries in .sift file */
|
293
|
+
/* */
|
294
|
+
/*************************************************************************/
|
295
|
+
|
296
|
+
|
297
|
+
void ProcessSift()
|
298
|
+
/* ----------- */
|
299
|
+
{
|
300
|
+
int E;
|
301
|
+
Attribute Att;
|
302
|
+
|
303
|
+
/* Find maximum discrete value */
|
304
|
+
|
305
|
+
ForEach(Att, 1, MaxAtt)
|
306
|
+
{
|
307
|
+
if ( Discrete(Att) && ! Exclude(Att) && MaxAttVal[Att] > MaxDiscrVal )
|
308
|
+
{
|
309
|
+
MaxDiscrVal = MaxAttVal[Att];
|
310
|
+
}
|
311
|
+
}
|
312
|
+
|
313
|
+
/* Allocate variables used globally */
|
314
|
+
|
315
|
+
LeftSS = Alloc((MaxDiscrVal>>3)+1, unsigned char);
|
316
|
+
Surprise = Alloc(MaxDiscrVal+1, float);
|
317
|
+
SomeNA = Alloc(MaxAtt+1, Boolean);
|
318
|
+
GLp = Alloc(101, CaseNo);
|
319
|
+
|
320
|
+
GEnv.Possible = Alloc(MaxDiscrVal+1, Boolean);
|
321
|
+
GEnv.Tested = Alloc(MaxAtt+1, int);
|
322
|
+
|
323
|
+
Caveat = Alloc(MaxAtt, CaveatRec);
|
324
|
+
ForEach(Att, 1, MaxAtt)
|
325
|
+
{
|
326
|
+
SomeNA[Att] = true;
|
327
|
+
Caveat[Att-1].Subset = Alloc((MaxDiscrVal>>3)+1, unsigned char);
|
328
|
+
}
|
329
|
+
|
330
|
+
/* Process successive entries */
|
331
|
+
|
332
|
+
while ( fscanf(Sf, "%d", &E) == 1 )
|
333
|
+
{
|
334
|
+
switch ( E )
|
335
|
+
{
|
336
|
+
case 0:
|
337
|
+
return;
|
338
|
+
|
339
|
+
case 1:
|
340
|
+
Case1();
|
341
|
+
break;
|
342
|
+
|
343
|
+
case 2:
|
344
|
+
Case2();
|
345
|
+
break;
|
346
|
+
|
347
|
+
case 3:
|
348
|
+
Case3();
|
349
|
+
break;
|
350
|
+
|
351
|
+
case 11:
|
352
|
+
Case11();
|
353
|
+
break;
|
354
|
+
|
355
|
+
case 12:
|
356
|
+
Case12();
|
357
|
+
break;
|
358
|
+
|
359
|
+
case 13:
|
360
|
+
Case13();
|
361
|
+
break;
|
362
|
+
|
363
|
+
case 21:
|
364
|
+
Case21();
|
365
|
+
break;
|
366
|
+
|
367
|
+
case 22:
|
368
|
+
Case22();
|
369
|
+
break;
|
370
|
+
|
371
|
+
default:
|
372
|
+
Error(BADSIFT, "entry", "");
|
373
|
+
}
|
374
|
+
}
|
375
|
+
}
|
376
|
+
|
377
|
+
|
378
|
+
|
379
|
+
/*************************************************************************/
|
380
|
+
/* */
|
381
|
+
/* Functions for each value of the switch variable */
|
382
|
+
/* */
|
383
|
+
/*************************************************************************/
|
384
|
+
|
385
|
+
|
386
|
+
void Case1()
|
387
|
+
/* ----- */
|
388
|
+
{
|
389
|
+
CaseNo i, Lp;
|
390
|
+
|
391
|
+
if ( fscanf(Sf, "%d\n", &ClassAtt) != 1 )
|
392
|
+
{
|
393
|
+
Error(BADSIFT, "1", "");
|
394
|
+
}
|
395
|
+
|
396
|
+
|
397
|
+
/* Remove cases with unknown value of new target and, if relevant,
|
398
|
+
those with values in excluded tails */
|
399
|
+
|
400
|
+
Lp = MaxCase;
|
401
|
+
for ( i = MaxCase ; i >= 0; i-- )
|
402
|
+
{
|
403
|
+
if ( Unknown(Case[i], ClassAtt) ||
|
404
|
+
( Continuous(ClassAtt) &&
|
405
|
+
( CVal(Case[i], ClassAtt) < LowTail[ClassAtt] ||
|
406
|
+
CVal(Case[i], ClassAtt) > HighTail[ClassAtt] ) ) )
|
407
|
+
{
|
408
|
+
Swap(i, Lp);
|
409
|
+
Lp--;
|
410
|
+
}
|
411
|
+
}
|
412
|
+
|
413
|
+
/* Initialise level etc */
|
414
|
+
|
415
|
+
GEnv.Level = -1;
|
416
|
+
GLp[0] = Lp;
|
417
|
+
}
|
418
|
+
|
419
|
+
|
420
|
+
|
421
|
+
void Case2()
|
422
|
+
/* ----- */
|
423
|
+
{
|
424
|
+
Attribute Att;
|
425
|
+
|
426
|
+
if ( fscanf(Sf, "%d\n", &Att) != 1 )
|
427
|
+
{
|
428
|
+
Error(BADSIFT, "2", "");
|
429
|
+
}
|
430
|
+
|
431
|
+
UseLogs[Att] = true;
|
432
|
+
}
|
433
|
+
|
434
|
+
|
435
|
+
|
436
|
+
void Case3()
|
437
|
+
/* ----- */
|
438
|
+
{
|
439
|
+
Attribute Att;
|
440
|
+
|
441
|
+
if ( fscanf(Sf, "%d", &Att) != 1 ||
|
442
|
+
fscanf(Sf, "%g %g\n", &LowTail[Att], &HighTail[Att]) != 2 )
|
443
|
+
{
|
444
|
+
Error(BADSIFT, "3", "");
|
445
|
+
}
|
446
|
+
}
|
447
|
+
|
448
|
+
|
449
|
+
|
450
|
+
void Case11()
|
451
|
+
/* ------ */
|
452
|
+
{
|
453
|
+
Attribute Att;
|
454
|
+
DiscrValue Br;
|
455
|
+
|
456
|
+
if ( fscanf(Sf, "%d %d %d\n", &GEnv.Level, &Att, &Br) != 3 )
|
457
|
+
{
|
458
|
+
Error(BADSIFT, "11", "");
|
459
|
+
}
|
460
|
+
|
461
|
+
Filter(Att, Br, 0, Nil);
|
462
|
+
}
|
463
|
+
|
464
|
+
|
465
|
+
|
466
|
+
void Case12()
|
467
|
+
/* ------ */
|
468
|
+
{
|
469
|
+
Attribute Att;
|
470
|
+
DiscrValue Br;
|
471
|
+
float Cut;
|
472
|
+
|
473
|
+
if ( fscanf(Sf, "%d %d %d %g\n", &GEnv.Level, &Att, &Br, &Cut) != 4 )
|
474
|
+
{
|
475
|
+
Error(BADSIFT, "12", "");
|
476
|
+
}
|
477
|
+
|
478
|
+
Filter(Att, Br, Cut, Nil);
|
479
|
+
}
|
480
|
+
|
481
|
+
|
482
|
+
|
483
|
+
void Case13()
|
484
|
+
/* ------ */
|
485
|
+
{
|
486
|
+
Attribute Att;
|
487
|
+
DiscrValue Br;
|
488
|
+
int Bytes, b, X;
|
489
|
+
|
490
|
+
if ( fscanf(Sf, "%d %d %d", &GEnv.Level, &Att, &Br) != 3 )
|
491
|
+
{
|
492
|
+
Error(BADSIFT, "13", "");
|
493
|
+
}
|
494
|
+
|
495
|
+
Bytes = (MaxAttVal[Att]>>3) + 1;
|
496
|
+
ForEach(b, 0, Bytes-1)
|
497
|
+
{
|
498
|
+
if ( fscanf(Sf, "%x", &X) != 1 )
|
499
|
+
{
|
500
|
+
Error(BADSIFT, "13+", "");
|
501
|
+
}
|
502
|
+
|
503
|
+
LeftSS[b] = X;
|
504
|
+
}
|
505
|
+
|
506
|
+
Filter(Att, Br, 0, LeftSS);
|
507
|
+
}
|
508
|
+
|
509
|
+
|
510
|
+
|
511
|
+
void Case21()
|
512
|
+
/* ------ */
|
513
|
+
{
|
514
|
+
CaseCount Cover, Anoms;
|
515
|
+
CaseNo i;
|
516
|
+
DiscrValue Expect, v;
|
517
|
+
float Frac;
|
518
|
+
Clust C=Nil;
|
519
|
+
|
520
|
+
if ( fscanf(Sf, "%d %g %d", &Cover, &Frac, &Expect) != 3 )
|
521
|
+
{
|
522
|
+
Error(BADSIFT, "21", "");
|
523
|
+
}
|
524
|
+
|
525
|
+
Anoms = rint(Cover * (1 - Frac));
|
526
|
+
|
527
|
+
ForEach(v, 1, MaxAttVal[ClassAtt])
|
528
|
+
{
|
529
|
+
Surprise[v] = 1;
|
530
|
+
}
|
531
|
+
|
532
|
+
while ( true )
|
533
|
+
{
|
534
|
+
if ( fscanf(Sf, "%d", &v) != 1 )
|
535
|
+
{
|
536
|
+
Error(BADSIFT, "21+", "");
|
537
|
+
}
|
538
|
+
|
539
|
+
if ( !v ) break;
|
540
|
+
|
541
|
+
if ( fscanf(Sf, "%g", &Surprise[v]) != 1 )
|
542
|
+
{
|
543
|
+
Error(BADSIFT, "21++", "");
|
544
|
+
}
|
545
|
+
}
|
546
|
+
|
547
|
+
ReadCaveats();
|
548
|
+
|
549
|
+
ForEach(i, 0, GLp[GEnv.Level+1])
|
550
|
+
{
|
551
|
+
v = XDVal(Case[i], ClassAtt);
|
552
|
+
|
553
|
+
if ( Surprise[v] < 1 && CheckCaveats(Case[i]) )
|
554
|
+
{
|
555
|
+
if ( ! C )
|
556
|
+
{
|
557
|
+
SetTestedAtts();
|
558
|
+
C = NewClust(Expect, 0.0, 0.0, Anoms, Cover);
|
559
|
+
}
|
560
|
+
|
561
|
+
FoundPossibleAnom(i, C, Surprise[v]);
|
562
|
+
}
|
563
|
+
}
|
564
|
+
}
|
565
|
+
|
566
|
+
|
567
|
+
|
568
|
+
void Case22()
|
569
|
+
/* ------ */
|
570
|
+
{
|
571
|
+
CaseCount Cover, Anoms;
|
572
|
+
CaseNo i;
|
573
|
+
Clust LowC=Nil, HighC=Nil;
|
574
|
+
float LowFrac, HighFrac, LowLim, HighLim, Mean, SD, Cv,
|
575
|
+
Z, LowLimZ, HighLimZ;
|
576
|
+
|
577
|
+
if ( fscanf(Sf, "%d %g %g %g %g %g %g",
|
578
|
+
&Cover, &Mean, &SD, &LowFrac, &LowLim, &HighFrac, &HighLim)
|
579
|
+
!= 7 )
|
580
|
+
{
|
581
|
+
Error(BADSIFT, "22", "");
|
582
|
+
}
|
583
|
+
|
584
|
+
if ( UseLogs[ClassAtt] )
|
585
|
+
{
|
586
|
+
LowLimZ = fabs(log(LowLim) - Mean) / SD;
|
587
|
+
HighLimZ = fabs(log(HighLim) - Mean) / SD;
|
588
|
+
}
|
589
|
+
else
|
590
|
+
{
|
591
|
+
LowLimZ = fabs(LowLim - Mean) / SD;
|
592
|
+
HighLimZ = fabs(HighLim - Mean) / SD;
|
593
|
+
}
|
594
|
+
|
595
|
+
ReadCaveats();
|
596
|
+
|
597
|
+
ForEach(i, 0, GLp[GEnv.Level+1])
|
598
|
+
{
|
599
|
+
Cv = CVal(Case[i], ClassAtt);
|
600
|
+
if ( UseLogs[ClassAtt] ) Cv = ( Cv > 0 ? log(Cv) : -1E38 );
|
601
|
+
|
602
|
+
Z = fabs(Mean - Cv) / SD;
|
603
|
+
|
604
|
+
if ( LowFrac > 0 && Cv < Mean &&
|
605
|
+
Z - LowLimZ >= MINABNORM - MAXNORM )
|
606
|
+
{
|
607
|
+
if ( CheckCaveats(Case[i]) )
|
608
|
+
{
|
609
|
+
if ( ! LowC )
|
610
|
+
{
|
611
|
+
SetTestedAtts();
|
612
|
+
Anoms = rint(Cover * (1 - LowFrac));
|
613
|
+
LowC = NewClust(Mean, SD, LowLim, Anoms, Cover);
|
614
|
+
}
|
615
|
+
|
616
|
+
FoundPossibleAnom(i, LowC, 1 / (Z * Z));
|
617
|
+
}
|
618
|
+
}
|
619
|
+
|
620
|
+
if ( HighFrac > 0 && Cv > Mean &&
|
621
|
+
Z - HighLimZ >= MINABNORM - MAXNORM )
|
622
|
+
{
|
623
|
+
if ( CheckCaveats(Case[i]) )
|
624
|
+
{
|
625
|
+
if ( ! HighC )
|
626
|
+
{
|
627
|
+
SetTestedAtts();
|
628
|
+
Anoms = rint(Cover * (1 - HighFrac));
|
629
|
+
HighC = NewClust(Mean, SD, HighLim, Anoms, Cover);
|
630
|
+
}
|
631
|
+
|
632
|
+
FoundPossibleAnom(i, HighC, 1 / (Z * Z));
|
633
|
+
}
|
634
|
+
}
|
635
|
+
}
|
636
|
+
}
|
637
|
+
|
638
|
+
|
639
|
+
|
640
|
+
/*************************************************************************/
|
641
|
+
/* */
|
642
|
+
/* Read any caveats on other attributes */
|
643
|
+
/* */
|
644
|
+
/*************************************************************************/
|
645
|
+
|
646
|
+
|
647
|
+
void ReadCaveats()
|
648
|
+
/* ----------- */
|
649
|
+
{
|
650
|
+
Attribute Att;
|
651
|
+
int Bytes, Byte, b;
|
652
|
+
|
653
|
+
NCaveat = 0;
|
654
|
+
|
655
|
+
while ( getc(Sf) == ' ' )
|
656
|
+
{
|
657
|
+
if ( fscanf(Sf, "%d", &Att) != 1 )
|
658
|
+
{
|
659
|
+
Error(BADSIFT, "caveat", "");
|
660
|
+
}
|
661
|
+
|
662
|
+
Caveat[NCaveat].Att = Att;
|
663
|
+
|
664
|
+
if ( Continuous(Att) )
|
665
|
+
{
|
666
|
+
if ( fscanf(Sf, "%g %g",
|
667
|
+
&Caveat[NCaveat].Low, &Caveat[NCaveat].High) != 2 )
|
668
|
+
{
|
669
|
+
Error(BADSIFT, "caveat", "");
|
670
|
+
}
|
671
|
+
}
|
672
|
+
else
|
673
|
+
{
|
674
|
+
Bytes = (MaxAttVal[Att]>>3) + 1;
|
675
|
+
ForEach(b, 0, Bytes-1)
|
676
|
+
{
|
677
|
+
if ( fscanf(Sf, "%x", &Byte) != 1 )
|
678
|
+
{
|
679
|
+
Error(BADSIFT, "caveat", "");
|
680
|
+
}
|
681
|
+
Caveat[NCaveat].Subset[b] = Byte;
|
682
|
+
}
|
683
|
+
}
|
684
|
+
|
685
|
+
NCaveat++;
|
686
|
+
}
|
687
|
+
}
|
688
|
+
|
689
|
+
|
690
|
+
|
691
|
+
/*************************************************************************/
|
692
|
+
/* */
|
693
|
+
/* Check that case satisfies all caveats */
|
694
|
+
/* */
|
695
|
+
/*************************************************************************/
|
696
|
+
|
697
|
+
|
698
|
+
Boolean CheckCaveats(Description Case)
|
699
|
+
/* ------------ */
|
700
|
+
{
|
701
|
+
Attribute Att;
|
702
|
+
int j;
|
703
|
+
ContValue Cv;
|
704
|
+
DiscrValue Dv;
|
705
|
+
|
706
|
+
for ( j = 0 ; j < NCaveat ; j++ )
|
707
|
+
{
|
708
|
+
Att = Caveat[j].Att;
|
709
|
+
|
710
|
+
if ( Continuous(Att) )
|
711
|
+
{
|
712
|
+
if ( ! Unknown(Case, Att) && ! NotApplic(Case, Att) &&
|
713
|
+
( (Cv = CVal(Case, Att)) < Caveat[j].Low ||
|
714
|
+
Cv > Caveat[j].High ) )
|
715
|
+
{
|
716
|
+
return false;
|
717
|
+
}
|
718
|
+
}
|
719
|
+
else
|
720
|
+
{
|
721
|
+
Dv = XDVal(Case, Att);
|
722
|
+
if ( In(Dv, Caveat[j].Subset) ) return false;
|
723
|
+
}
|
724
|
+
}
|
725
|
+
|
726
|
+
return true;
|
727
|
+
}
|
728
|
+
|
729
|
+
|
730
|
+
|
731
|
+
/*************************************************************************/
|
732
|
+
/* */
|
733
|
+
/* Found one -- see whether more interesting than current */
|
734
|
+
/* */
|
735
|
+
/*************************************************************************/
|
736
|
+
|
737
|
+
|
738
|
+
void FoundPossibleAnom(CaseNo i, Clust C, float Xv)
|
739
|
+
/* ----------------- */
|
740
|
+
{
|
741
|
+
Clust OldC;
|
742
|
+
|
743
|
+
OldC = OutClust(Case[i]);
|
744
|
+
|
745
|
+
if ( ! OldC ||
|
746
|
+
C->NCond < OldC->NCond ||
|
747
|
+
C->NCond == OldC->NCond && Xv < OutXVal(Case[i]) )
|
748
|
+
{
|
749
|
+
RecordOutlier(i, C, Xv);
|
750
|
+
}
|
751
|
+
}
|
752
|
+
|
753
|
+
|
754
|
+
|
755
|
+
/*************************************************************************/
|
756
|
+
/* */
|
757
|
+
/* Add test to the test stack and select relevant cases */
|
758
|
+
/* */
|
759
|
+
/*************************************************************************/
|
760
|
+
|
761
|
+
|
762
|
+
void Filter(Attribute Att, DiscrValue Br, ContValue Cut, Set Left)
|
763
|
+
/* ------ */
|
764
|
+
{
|
765
|
+
NoteTest(Att, Br, Cut, Left);
|
766
|
+
|
767
|
+
GLp[GEnv.Level+1] = Group(Att, Br, 0, GLp[GEnv.Level], Cut, Left) - 1;
|
768
|
+
}
|
769
|
+
|
770
|
+
|
771
|
+
|
772
|
+
/*************************************************************************/
|
773
|
+
/* */
|
774
|
+
/* Determine attributes used in all conditions */
|
775
|
+
/* */
|
776
|
+
/*************************************************************************/
|
777
|
+
|
778
|
+
|
779
|
+
void SetTestedAtts()
|
780
|
+
/* ------------- */
|
781
|
+
{
|
782
|
+
Attribute Att;
|
783
|
+
int i;
|
784
|
+
|
785
|
+
ForEach(Att, 1, MaxAtt)
|
786
|
+
{
|
787
|
+
GEnv.Tested[Att] = false;
|
788
|
+
}
|
789
|
+
|
790
|
+
ForEach(i, 0, GEnv.Level)
|
791
|
+
{
|
792
|
+
GEnv.Tested[GEnv.Test[i].Att] = true;
|
793
|
+
}
|
794
|
+
}
|