DataSet::DataSet(bool classAtStart, string file, double ratio)
{
name = file;
uniqueClasses=0;
//int classC=0;
classes = new vector<string>;
data = new vector< vector<double>* >;
ifstream ifs(file.c_str());
if(!ifs.good())
return;
char buf[4096];
string strBuf;
string * tmpStr;
while(!ifs.eof()){
ifs.getline(buf,sizeof buf);
strBuf = buf;
if(strBuf.length()==0)
break;
if(classAtStart){
if(!isIn(classes,*getNext(strBuf)))
uniqueClasses++;
classes->push_back(*getNext(strBuf));
strBuf = getRest(strBuf);
}else{
if(!isIn(classes,*getLast(strBuf)))
uniqueClasses++;
classes->push_back(*getLast(strBuf));
strBuf = getRestEnd(strBuf);
}
vector<double> * intvector = new vector<double>;
while((tmpStr = getNext(strBuf)) != 0){
intvector->push_back(atof(tmpStr->c_str()));
strBuf = getRest(strBuf);
}
data->push_back(intvector);
}
ifs.close();
cerr << "read "<<data->size()<<" datapoints, with "<<data->at(0)->size()<<" paramaters classes: "<<classes->size()<<" unique:"<<uniqueClasses<<"\n";
tests = (int)(data->size()*ratio);
trainings = data->size()-tests;
Dummy myDummy;
int ** tmpCoord = myDummy.randomCoords(data->size(),0);
trainCoords = new int [trainings];
testCoords = new int [tests];
int c = 0;
for(;c<tests;c++)
testCoords[c] = tmpCoord[c][0];
for(;c<(tests+trainings);c++)
trainCoords[(c-tests)] = tmpCoord[c][0];
cerr << "tests: "<<tests<<" trainings: "<<trainings<<endl;
}
// Get arguments according to specified format
bool LineParser::getArgsFormatted(ParseFormat& format, int optionMask, bool readLine)
{
Messenger::enter("LineParser::getArgsFormatted");
bool done = false;
int result;
arguments_.clear();
// Get line from file?
if (readLine)
{
// Returns : 0=ok, 1=error, -1=eof
result = readNextLine(optionMask);
if (result != 0)
{
Messenger::exit("LineParser::getArgsFormatted");
return result;
}
}
// Loop over chunks in format
QString arg;
bool failed = false;
for (ParseChunk* chunk = format.chunks(); chunk != NULL; chunk = chunk->next)
{
switch (chunk->type())
{
case (ParseChunk::DelimitedChunk):
if (getNextArg(optionMask, arg)) arguments_ << arg;
else failed = true;
break;
case (ParseChunk::FormattedChunk):
if (getNextN(optionMask, chunk->formatLength(), arg)) arguments_ << arg;
else failed = true;
break;
case (ParseChunk::DiscardChunk):
if (!getNextN(optionMask, chunk->formatLength(), arg)) failed = true;
break;
case (ParseChunk::PlainTextChunk):
if (!getNextN(optionMask, chunk->cFormat().length(), arg)) failed = true;
break;
case (ParseChunk::GreedyDelimitedChunk):
if (getRest(arg)) arguments_ << arg;
else failed = true;
break;
default:
break;
}
if (failed) break;
}
Messenger::exit("LineParser::getArgsFormatted");
return 0;
}