int main (/*int argc, char *argv[]*/) {
QTextCodec::setCodecForCStrings(QTextCodec::codecForName("koi8-u"));
QTextCodec::setCodecForLocale(QTextCodec::codecForName("koi8-u"));
/*
const char *str = "\"this is a \\u04e3 test \\x21 string \\120!\\n\"";
int bp = 0;
qDebug() << "strlen:" << strlen(str);
qDebug() << "str:" << str;
qDebug() << K8JSON::stringLength(str, strlen(str), &bp);
qDebug() << " " << bp;
//bool parseString (QString &str, const char *s, int charCount, int *bytesProcessed=0);
*/
testCounter();
testReader();
testReaderAll();
testReaderAll2();
return 0;
}
int main(int argc, char *argv[]) {
std::string usage = "Usage: runClassifier [options] optionFile testFile numTrainingInstances";
parseCommandLineArgs(&argc,&argv,usage,3,3);
std::string optionFile = argv[1];
std::string testFile = argv[2];
unsigned int numTrainingInstances = boost::lexical_cast<unsigned int>(argv[3]);
Json::Value options;
if (! readJson(optionFile,options))
return 1;
ClassifierPtr classifier = createClassifier(options);
std::ifstream testIn(testFile.c_str());
ArffReader testReader(testIn);
for (unsigned int count = 0; count < numTrainingInstances; count++) {
if (testReader.isDone()) {
std::cerr << "ERROR: Insufficient training data" << std::endl;
std::cerr << "Found " << count << " instances, expected " << numTrainingInstances << std::endl;
exit(2);
}
InstancePtr instance = testReader.next();
classifier->addData(instance);
}
double trainingTime = 0.0;
if (FLAGS_train) {
double startTime = getTime();
classifier->train(false);
trainingTime = getTime() - startTime;
std::cout << "Training time: " << trainingTime << std::endl;
}
std::cout << "------------------------------------------" << std::endl;
//std::cout << *classifier << std::endl;
double correct = 0.0;
int correctCount = 0;
int count;
for (count = 0; !testReader.isDone(); count++) {
InstancePtr instance = testReader.next();
Classification c;
classifier->classify(instance,c);
//std::cout << *instance << std::endl;
//std::cout << " ";
//for (unsigned int i = 0; i < c.size(); i++)
//std::cout << c[i] << " ";
//std::cout << std::endl;
// calculate the fraction correct
correct += c[instance->label];
// calculate whether most probable was correct
unsigned int maxInd = vectorMaxInd(c);
if (maxInd == instance->label)
correctCount++;
}
testIn.close();
std::cout << "------------------------------------------" << std::endl;
std::cout << "Target Training Insts: " << numTrainingInstances << std::endl;
std::cout << "Testing Insts: " << count << std::endl;
std::cout << "Frac Correct: " << correct << "(" << correct / count << ")" << std::endl;
std::cout << "Num Correct: " << correctCount << "(" << correctCount / (float)count << ")" << std::endl;
std::cout << "Training time: " << trainingTime << std::endl;
return 0;
}
