1 : |
agomez |
1 |
#include "sionstep.h" |
2 : |
|
|
#include "areatime.h" |
3 : |
|
|
#include "errorhandler.h" |
4 : |
|
|
#include "readfunc.h" |
5 : |
|
|
#include "readword.h" |
6 : |
|
|
#include "gadget.h" |
7 : |
|
|
#include "global.h" |
8 : |
|
|
|
9 : |
|
|
SIOnStep::~SIOnStep() {
|
10 : |
|
|
int i;
|
11 : |
|
|
if (LgrpDiv != 0)
|
12 : |
|
|
delete LgrpDiv;
|
13 : |
|
|
if (LR != 0)
|
14 : |
|
|
delete LR;
|
15 : |
|
|
for (i = 0; i < areaindex.Size(); i++)
|
16 : |
|
|
delete[] areaindex[i];
|
17 : |
|
|
for (i = 0; i < colindex.Size(); i++)
|
18 : |
|
|
delete[] colindex[i];
|
19 : |
|
|
for (i = 0; i < obsIndex.Size(); i++) {
|
20 : |
|
|
delete obsIndex[i];
|
21 : |
|
|
delete modelIndex[i];
|
22 : |
|
|
}
|
23 : |
|
|
for (i = 0; i < weightIndex.Size(); i++)
|
24 : |
|
|
delete weightIndex[i];
|
25 : |
|
|
}
|
26 : |
|
|
|
27 : |
|
|
SIOnStep::SIOnStep(CommentStream& infile, const char* datafilename, const CharPtrVector& aindex,
|
28 : |
|
|
const TimeClass* const TimeInfo, const IntMatrix& areas, const CharPtrVector& charindex,
|
29 : |
|
|
const char* givenname, int bio, SIType type) : HasName(givenname), Areas(areas), alptr(0) {
|
30 : |
|
|
|
31 : |
|
|
biomass = bio;
|
32 : |
|
|
sitype = type;
|
33 : |
|
|
useweight = 0;
|
34 : |
|
|
timeindex = 0;
|
35 : |
|
|
slope = 0.0;
|
36 : |
|
|
intercept = 0.0;
|
37 : |
|
|
|
38 : |
|
|
int i;
|
39 : |
|
|
for (i = 0; i < aindex.Size(); i++) {
|
40 : |
|
|
areaindex.resize(new char[strlen(aindex[i]) + 1]);
|
41 : |
|
|
strcpy(areaindex[i], aindex[i]);
|
42 : |
|
|
}
|
43 : |
|
|
for (i = 0; i < charindex.Size(); i++) {
|
44 : |
|
|
colindex.resize(new char[strlen(charindex[i]) + 1]);
|
45 : |
|
|
strcpy(colindex[i], charindex[i]);
|
46 : |
|
|
}
|
47 : |
|
|
|
48 : |
|
|
//read the fittype
|
49 : |
|
|
readSIRegressionData(infile);
|
50 : |
|
|
|
51 : |
|
|
//read the survey indices data from the datafile
|
52 : |
|
|
ifstream datafile;
|
53 : |
|
|
CommentStream subdata(datafile);
|
54 : |
|
|
datafile.open(datafilename, ios::in);
|
55 : |
|
|
handle.checkIfFailure(datafile, datafilename);
|
56 : |
|
|
handle.Open(datafilename);
|
57 : |
|
|
readSIData(subdata, TimeInfo);
|
58 : |
|
|
handle.Close();
|
59 : |
|
|
datafile.close();
|
60 : |
|
|
datafile.clear();
|
61 : |
|
|
|
62 : |
|
|
//resize to store the regression information
|
63 : |
|
|
slopes.AddRows(areaindex.Size(), colindex.Size(), slope);
|
64 : |
|
|
intercepts.AddRows(areaindex.Size(), colindex.Size(), intercept);
|
65 : |
|
|
sse.AddRows(areaindex.Size(), colindex.Size(), 0.0);
|
66 : |
|
|
if (useweight)
|
67 : |
|
|
tmpWeight.resize(Years.Size(), 0.0);
|
68 : |
|
|
tmpModel.resize(Years.Size(), 0.0);
|
69 : |
|
|
tmpData.resize(Years.Size(), 0.0);
|
70 : |
|
|
likelihoodValues.resize(areaindex.Size(), 0.0);
|
71 : |
|
|
}
|
72 : |
|
|
|
73 : |
|
|
void SIOnStep::readSIData(CommentStream& infile, const TimeClass* const TimeInfo) {
|
74 : |
|
|
|
75 : |
|
|
double tmpnumber, tmpweight;
|
76 : |
|
|
char tmparea[MaxStrLength], tmplabel[MaxStrLength];
|
77 : |
|
|
strncpy(tmparea, "", MaxStrLength);
|
78 : |
|
|
strncpy(tmplabel, "", MaxStrLength);
|
79 : |
|
|
int keepdata, timeid, colid, areaid;
|
80 : |
|
|
int i, year, step, count, reject;
|
81 : |
|
|
|
82 : |
|
|
//Check the number of columns in the inputfile
|
83 : |
|
|
infile >> ws;
|
84 : |
|
|
if ((!useweight) && (countColumns(infile) != 5))
|
85 : |
|
|
handle.logFileMessage(LOGFAIL, "wrong number of columns in inputfile - should be 5");
|
86 : |
|
|
else if ((useweight) && (countColumns(infile) != 6))
|
87 : |
|
|
handle.logFileMessage(LOGFAIL, "wrong number of columns in inputfile - should be 6");
|
88 : |
|
|
|
89 : |
|
|
year = step = count = reject = 0;
|
90 : |
|
|
while (!infile.eof()) {
|
91 : |
|
|
keepdata = 1;
|
92 : |
|
|
|
93 : |
|
|
if (useweight)
|
94 : |
|
|
infile >> year >> step >> tmparea >> tmplabel >> tmpnumber >> tmpweight >> ws;
|
95 : |
|
|
else
|
96 : |
|
|
infile >> year >> step >> tmparea >> tmplabel >> tmpnumber >> ws;
|
97 : |
|
|
|
98 : |
|
|
//crude check to see if something has gone wrong and avoid infinite loops
|
99 : |
|
|
if (strlen(tmparea) == 0)
|
100 : |
|
|
handle.logFileMessage(LOGFAIL, "failed to read data from file");
|
101 : |
|
|
|
102 : |
|
|
//if tmparea is in areaindex keep data, else dont keep the data
|
103 : |
|
|
areaid = -1;
|
104 : |
|
|
for (i = 0; i < areaindex.Size(); i++)
|
105 : |
|
|
if (strcasecmp(areaindex[i], tmparea) == 0)
|
106 : |
|
|
areaid = i;
|
107 : |
|
|
|
108 : |
|
|
if (areaid == -1)
|
109 : |
|
|
keepdata = 0;
|
110 : |
|
|
|
111 : |
|
|
//if tmplabel is in colindex find colid, else dont keep the data
|
112 : |
|
|
colid = -1;
|
113 : |
|
|
for (i = 0; i < colindex.Size(); i++)
|
114 : |
|
|
if (strcasecmp(colindex[i], tmplabel) == 0)
|
115 : |
|
|
colid = i;
|
116 : |
|
|
|
117 : |
|
|
if (colid == -1)
|
118 : |
|
|
keepdata = 0;
|
119 : |
|
|
|
120 : |
|
|
//check if the year and step are in the simulation
|
121 : |
|
|
timeid = -1;
|
122 : |
|
|
if ((TimeInfo->isWithinPeriod(year, step)) && (keepdata == 1)) {
|
123 : |
|
|
for (i = 0; i < Years.Size(); i++)
|
124 : |
|
|
if ((Years[i] == year) && (Steps[i] == step))
|
125 : |
|
|
timeid = i;
|
126 : |
|
|
|
127 : |
|
|
if (timeid == -1) {
|
128 : |
|
|
Years.resize(1, year);
|
129 : |
|
|
Steps.resize(1, step);
|
130 : |
|
|
timeid = (Years.Size() - 1);
|
131 : |
|
|
obsIndex.resize(new DoubleMatrix(areaindex.Size(), colindex.Size(), 0.0));
|
132 : |
|
|
modelIndex.resize(new DoubleMatrix(areaindex.Size(), colindex.Size(), 0.0));
|
133 : |
|
|
if (useweight)
|
134 : |
|
|
weightIndex.resize(new DoubleMatrix(areaindex.Size(), colindex.Size(), 0.0));
|
135 : |
|
|
}
|
136 : |
|
|
|
137 : |
|
|
} else
|
138 : |
|
|
keepdata = 0;
|
139 : |
|
|
|
140 : |
|
|
if (keepdata == 1) {
|
141 : |
|
|
//survey indices data is required, so store it
|
142 : |
|
|
count++;
|
143 : |
|
|
(*obsIndex[timeid])[areaid][colid] = tmpnumber;
|
144 : |
|
|
if (useweight)
|
145 : |
|
|
(*weightIndex[timeid])[areaid][colid] = tmpweight;
|
146 : |
|
|
} else
|
147 : |
|
|
reject++; //count number of rejected data points read from file
|
148 : |
|
|
}
|
149 : |
|
|
|
150 : |
|
|
AAT.addActions(Years, Steps, TimeInfo);
|
151 : |
|
|
if (count == 0)
|
152 : |
|
|
handle.logMessage(LOGWARN, "Warning in surveyindex - found no data in the data file for", this->getName());
|
153 : |
|
|
if (Years.Size() < 2)
|
154 : |
|
|
handle.logMessage(LOGWARN, "Warning in surveyindex - insufficient data to fit a regression line for", this->getName());
|
155 : |
|
|
|
156 : |
|
|
if ((handle.getLogLevel() >= LOGWARN) && (Steps.Size() > 0) && (this->getSIType() != SIEFFORT)) {
|
157 : |
|
|
//JMB to be comparable, this should only take place on the same step in each year
|
158 : |
|
|
//However the effort data will be on most timesteps so this check isnt required
|
159 : |
|
|
step = Steps[0];
|
160 : |
|
|
timeid = 0;
|
161 : |
|
|
for (i = 1; i < Steps.Size(); i++)
|
162 : |
|
|
if (Steps[i] != step)
|
163 : |
|
|
timeid++;
|
164 : |
|
|
|
165 : |
|
|
if (timeid != 0)
|
166 : |
|
|
handle.logMessage(LOGWARN, "Warning in surveyindex - differing timesteps for", this->getName());
|
167 : |
|
|
}
|
168 : |
|
|
|
169 : |
|
|
if (reject != 0)
|
170 : |
|
|
handle.logMessage(LOGMESSAGE, "Discarded invalid surveyindex data - number of invalid entries", reject);
|
171 : |
|
|
handle.logMessage(LOGMESSAGE, "Read surveyindex data file - number of entries", count);
|
172 : |
|
|
}
|
173 : |
|
|
|
174 : |
|
|
void SIOnStep::readSIRegressionData(CommentStream& infile) {
|
175 : |
|
|
|
176 : |
|
|
char text[MaxStrLength];
|
177 : |
|
|
strncpy(text, "", MaxStrLength);
|
178 : |
|
|
|
179 : |
|
|
infile >> ws >> text;
|
180 : |
|
|
if (strcasecmp(text, "linearfit") == 0) {
|
181 : |
|
|
fittype = LINEARFIT;
|
182 : |
|
|
LR = new LinearRegression(FREE);
|
183 : |
|
|
} else if (strcasecmp(text, "loglinearfit") == 0) {
|
184 : |
|
|
fittype = LOGLINEARFIT;
|
185 : |
|
|
LR = new LogLinearRegression(FREE);
|
186 : |
|
|
} else if (strcasecmp(text, "weightlinearfit") == 0) {
|
187 : |
|
|
useweight = 1;
|
188 : |
|
|
fittype = WEIGHTLINEARFIT;
|
189 : |
|
|
LR = new WeightRegression(FREE);
|
190 : |
|
|
} else if (strcasecmp(text, "logweightlinearfit") == 0) {
|
191 : |
|
|
useweight = 1;
|
192 : |
|
|
fittype = LOGWEIGHTLINEARFIT;
|
193 : |
|
|
LR = new LogWeightRegression(FREE);
|
194 : |
|
|
} else if (strcasecmp(text, "fixedslopelinearfit") == 0) {
|
195 : |
|
|
fittype = FIXEDSLOPELINEARFIT;
|
196 : |
|
|
LR = new LinearRegression(FIXEDSLOPE);
|
197 : |
|
|
} else if (strcasecmp(text, "fixedslopeloglinearfit") == 0) {
|
198 : |
|
|
fittype = FIXEDSLOPELOGLINEARFIT;
|
199 : |
|
|
LR = new LogLinearRegression(FIXEDSLOPE);
|
200 : |
|
|
} else if (strcasecmp(text, "fixedslopeweightlinearfit") == 0) {
|
201 : |
|
|
useweight = 1;
|
202 : |
|
|
fittype = FIXEDSLOPEWEIGHTLINEARFIT;
|
203 : |
|
|
LR = new WeightRegression(FIXEDSLOPE);
|
204 : |
|
|
} else if (strcasecmp(text, "fixedslopelogweightlinearfit") == 0) {
|
205 : |
|
|
useweight = 1;
|
206 : |
|
|
fittype = FIXEDSLOPELOGWEIGHTLINEARFIT;
|
207 : |
|
|
LR = new LogWeightRegression(FIXEDSLOPE);
|
208 : |
|
|
} else if (strcasecmp(text, "fixedinterceptlinearfit") == 0) {
|
209 : |
|
|
fittype = FIXEDINTERCEPTLINEARFIT;
|
210 : |
|
|
LR = new LinearRegression(FIXEDINTERCEPT);
|
211 : |
|
|
} else if (strcasecmp(text, "fixedinterceptloglinearfit") == 0) {
|
212 : |
|
|
fittype = FIXEDINTERCEPTLOGLINEARFIT;
|
213 : |
|
|
LR = new LogLinearRegression(FIXEDINTERCEPT);
|
214 : |
|
|
} else if (strcasecmp(text, "fixedinterceptweightlinearfit") == 0) {
|
215 : |
|
|
useweight = 1;
|
216 : |
|
|
fittype = FIXEDINTERCEPTWEIGHTLINEARFIT;
|
217 : |
|
|
LR = new WeightRegression(FIXEDINTERCEPT);
|
218 : |
|
|
} else if (strcasecmp(text, "fixedinterceptlogweightlinearfit") == 0) {
|
219 : |
|
|
useweight = 1;
|
220 : |
|
|
fittype = FIXEDINTERCEPTLOGWEIGHTLINEARFIT;
|
221 : |
|
|
LR = new LogWeightRegression(FIXEDINTERCEPT);
|
222 : |
|
|
} else if (strcasecmp(text, "fixedlinearfit") == 0) {
|
223 : |
|
|
fittype = FIXEDLINEARFIT;
|
224 : |
|
|
LR = new LinearRegression(FIXED);
|
225 : |
|
|
} else if (strcasecmp(text, "fixedloglinearfit") == 0) {
|
226 : |
|
|
fittype = FIXEDLOGLINEARFIT;
|
227 : |
|
|
LR = new LogLinearRegression(FIXED);
|
228 : |
|
|
} else if (strcasecmp(text, "fixedweightlinearfit") == 0) {
|
229 : |
|
|
useweight = 1;
|
230 : |
|
|
fittype = FIXEDWEIGHTLINEARFIT;
|
231 : |
|
|
LR = new WeightRegression(FIXED);
|
232 : |
|
|
} else if (strcasecmp(text, "fixedlogweightlinearfit") == 0) {
|
233 : |
|
|
useweight = 1;
|
234 : |
|
|
fittype = FIXEDLOGWEIGHTLINEARFIT;
|
235 : |
|
|
LR = new LogWeightRegression(FIXED);
|
236 : |
|
|
} else
|
237 : |
|
|
handle.logFileMessage(LOGFAIL, "\nError in surveyindex - unrecognised fittype", text);
|
238 : |
|
|
|
239 : |
|
|
switch (fittype) {
|
240 : |
|
|
case LINEARFIT:
|
241 : |
|
|
case LOGLINEARFIT:
|
242 : |
|
|
case WEIGHTLINEARFIT:
|
243 : |
|
|
case LOGWEIGHTLINEARFIT:
|
244 : |
|
|
break;
|
245 : |
|
|
case FIXEDSLOPELINEARFIT:
|
246 : |
|
|
case FIXEDSLOPELOGLINEARFIT:
|
247 : |
|
|
case FIXEDSLOPEWEIGHTLINEARFIT:
|
248 : |
|
|
case FIXEDSLOPELOGWEIGHTLINEARFIT:
|
249 : |
|
|
readWordAndVariable(infile, "slope", slope);
|
250 : |
|
|
LR->setSlope(slope);
|
251 : |
|
|
break;
|
252 : |
|
|
case FIXEDINTERCEPTLINEARFIT:
|
253 : |
|
|
case FIXEDINTERCEPTLOGLINEARFIT:
|
254 : |
|
|
case FIXEDINTERCEPTWEIGHTLINEARFIT:
|
255 : |
|
|
case FIXEDINTERCEPTLOGWEIGHTLINEARFIT:
|
256 : |
|
|
readWordAndVariable(infile, "intercept", intercept);
|
257 : |
|
|
LR->setIntercept(intercept);
|
258 : |
|
|
break;
|
259 : |
|
|
case FIXEDLINEARFIT:
|
260 : |
|
|
case FIXEDLOGLINEARFIT:
|
261 : |
|
|
case FIXEDWEIGHTLINEARFIT:
|
262 : |
|
|
case FIXEDLOGWEIGHTLINEARFIT:
|
263 : |
|
|
readWordAndVariable(infile, "slope", slope);
|
264 : |
|
|
readWordAndVariable(infile, "intercept", intercept);
|
265 : |
|
|
LR->setSlope(slope);
|
266 : |
|
|
LR->setIntercept(intercept);
|
267 : |
|
|
break;
|
268 : |
|
|
default:
|
269 : |
|
|
handle.logFileMessage(LOGFAIL, "\nError in surveyindex - unrecognised fittype", text);
|
270 : |
|
|
break;
|
271 : |
|
|
}
|
272 : |
|
|
|
273 : |
|
|
//JMB - check that the slope of the regression line is positive
|
274 : |
|
|
if (slope < 0.0)
|
275 : |
|
|
handle.logFileMessage(LOGFAIL, "\nError in surveyindex - slope of the regression line must be positive", slope);
|
276 : |
|
|
}
|
277 : |
|
|
|
278 : |
|
|
void SIOnStep::printLikelihood(ofstream& outfile, const TimeClass* const TimeInfo) {
|
279 : |
|
|
|
280 : |
|
|
int a, i;
|
281 : |
|
|
if (AAT.atCurrentTime(TimeInfo)) {
|
282 : |
|
|
timeindex = -1;
|
283 : |
|
|
for (i = 0; i < Years.Size(); i++)
|
284 : |
|
|
if ((Years[i] == TimeInfo->getYear()) && (Steps[i] == TimeInfo->getStep()))
|
285 : |
|
|
timeindex = i;
|
286 : |
|
|
if (timeindex == -1)
|
287 : |
|
|
handle.logMessage(LOGFAIL, "Error in surveyindex - invalid timestep");
|
288 : |
|
|
|
289 : |
|
|
for (a = 0; a < areaindex.Size(); a++) {
|
290 : |
|
|
for (i = 0; i < colindex.Size(); i++) {
|
291 : |
|
|
outfile << setw(lowwidth) << Years[timeindex] << sep << setw(lowwidth)
|
292 : |
|
|
<< Steps[timeindex] << sep << setw(printwidth) << areaindex[a] << sep
|
293 : |
|
|
<< setw(printwidth) << colindex[i] << sep << setw(largewidth);
|
294 : |
|
|
|
295 : |
|
|
//JMB crude filter to remove the 'silly' values from the output
|
296 : |
|
|
if ((*modelIndex[timeindex])[a][i] < rathersmall)
|
297 : |
|
|
outfile << 0;
|
298 : |
|
|
else
|
299 : |
|
|
outfile << setprecision(largeprecision) << (*modelIndex[timeindex])[a][i];
|
300 : |
|
|
|
301 : |
|
|
if (useweight)
|
302 : |
|
|
outfile << sep << setw(printwidth) << (*weightIndex[timeindex])[a][i];
|
303 : |
|
|
outfile << endl;
|
304 : |
|
|
}
|
305 : |
|
|
}
|
306 : |
|
|
}
|
307 : |
|
|
|
308 : |
|
|
//JMB - this is nasty hack to output the regression information
|
309 : |
|
|
if (TimeInfo->getTime() == TimeInfo->numTotalSteps()) {
|
310 : |
|
|
for (a = 0; a < areaindex.Size(); a++) {
|
311 : |
|
|
outfile << "; Regression information for area " << areaindex[a] << endl;
|
312 : |
|
|
for (i = 0; i < colindex.Size(); i++)
|
313 : |
|
|
outfile << "; " << colindex[i] << " intercept " << intercepts[a][i]
|
314 : |
|
|
<< " slope " << slopes[a][i] << " sse " << sse[a][i] << endl;
|
315 : |
|
|
}
|
316 : |
|
|
}
|
317 : |
|
|
}
|
318 : |
|
|
|
319 : |
|
|
double SIOnStep::calcSSE() {
|
320 : |
|
|
|
321 : |
|
|
if (handle.getLogLevel() >= LOGMESSAGE)
|
322 : |
|
|
handle.logMessage(LOGMESSAGE, "Calculating likelihood score for surveyindex component", this->getName());
|
323 : |
|
|
|
324 : |
|
|
int a, i, j;
|
325 : |
|
|
double score = 0.0;
|
326 : |
|
|
for (a = 0; a < areaindex.Size(); a++) {
|
327 : |
|
|
likelihoodValues[a] = 0.0;
|
328 : |
|
|
for (i = 0; i < colindex.Size(); i++) {
|
329 : |
|
|
for (j = 0; j < tmpData.Size(); j++) {
|
330 : |
|
|
tmpData[j] = (*obsIndex[j])[a][i];
|
331 : |
|
|
tmpModel[j] = (*modelIndex[j])[a][i];
|
332 : |
|
|
}
|
333 : |
|
|
|
334 : |
|
|
//if the weights are required then pass them to the regression line
|
335 : |
|
|
if (useweight) {
|
336 : |
|
|
for (j = 0; j < tmpWeight.Size(); j++)
|
337 : |
|
|
tmpWeight[j] = (*weightIndex[j])[a][i];
|
338 : |
|
|
LR->setWeights(tmpWeight);
|
339 : |
|
|
}
|
340 : |
|
|
|
341 : |
|
|
//fit the data to the (log) linear regression line
|
342 : |
|
|
LR->storeVectors(tmpModel, tmpData);
|
343 : |
|
|
LR->calcFit();
|
344 : |
|
|
|
345 : |
|
|
//and then store the results
|
346 : |
|
|
slopes[a][i] = LR->getSlope();
|
347 : |
|
|
intercepts[a][i] = LR->getIntercept();
|
348 : |
|
|
sse[a][i] = LR->getSSE();
|
349 : |
|
|
likelihoodValues[a] += LR->getSSE();
|
350 : |
|
|
}
|
351 : |
|
|
score += likelihoodValues[a];
|
352 : |
|
|
}
|
353 : |
|
|
|
354 : |
|
|
if (handle.getLogLevel() >= LOGMESSAGE)
|
355 : |
|
|
handle.logMessage(LOGMESSAGE, "The likelihood score from the regression line for this component is", score);
|
356 : |
|
|
return score;
|
357 : |
|
|
}
|
358 : |
|
|
|
359 : |
|
|
void SIOnStep::printSummary(ofstream& outfile, const double weight) {
|
360 : |
|
|
int a;
|
361 : |
|
|
for (a = 0; a < areaindex.Size(); a++)
|
362 : |
|
|
outfile << "all all " << setw(printwidth) << areaindex[a] << sep
|
363 : |
|
|
<< setw(largewidth) << this->getName() << sep << setprecision(smallprecision)
|
364 : |
|
|
<< setw(smallwidth) << weight << sep << setprecision(largeprecision)
|
365 : |
|
|
<< setw(largewidth) << likelihoodValues[a] << endl;
|
366 : |
|
|
outfile.flush();
|
367 : |
|
|
}
|
368 : |
|
|
|
369 : |
|
|
void SIOnStep::Reset() {
|
370 : |
|
|
int i;
|
371 : |
|
|
for (i = 0; i < modelIndex.Size(); i++)
|
372 : |
|
|
(*modelIndex[i]).setToZero();
|
373 : |
|
|
}
|