void GBDT::LogLossProcess(DataVector *d, size_t samples, int i) {
#ifdef USE_OPENMP
#pragma omp parallel for
#endif
for (size_t j = 0; j < samples; ++j) {
ValueType p = Predict(*(*d)[j], i);
(*d)[j]->target =
static_cast<ValueType>(LogitLossGradient((*d)[j]->label, p));
}
if (g_conf.debug) {
Auc auc;
DataVector::iterator iter = d->begin();
for ( ; iter != d->end(); ++iter) {
ValueType p = Logit(Predict(**iter, i));
auc.Add(p, (*iter)->label);
}
std::cout << "auc: " << auc.CalculateAuc() << std::endl;
}
}
int main(int argc, char *argv[]) {
std::srand ( unsigned ( std::time(0) ) );
#ifdef USE_OPENMP
const int threads_wanted = 4;
omp_set_num_threads(threads_wanted);
#endif
g_conf.number_of_feature = 79;
g_conf.max_depth = 6;
g_conf.iterations = 10;
g_conf.shrinkage = 0.1F;
if (argc < 3) return -1;
std::string train_file(argv[1]);
std::string test_file(argv[2]);
if (argc > 3) {
g_conf.max_depth = boost::lexical_cast<int>(argv[3]);
}
if (argc > 4) {
g_conf.iterations = boost::lexical_cast<int>(argv[4]);
}
if (argc > 5) {
g_conf.shrinkage = boost::lexical_cast<float>(argv[5]);
}
if (argc > 6) {
g_conf.feature_sample_ratio = boost::lexical_cast<float>(argv[6]);
}
if (argc > 7) {
g_conf.data_sample_ratio = boost::lexical_cast<float>(argv[7]);
}
int debug = 0;
if (argc > 8) {
debug = boost::lexical_cast<int>(argv[8]);
}
g_conf.loss = LOG_LIKELIHOOD;
g_conf.debug = debug > 0? true : false;
DataVector d;
bool r = LoadDataFromFile(train_file, &d);
assert(r);
g_conf.min_leaf_size = d.size() / 40;
std::cout << "configure: " << std::endl
<< g_conf.ToString() << std::endl;
if (argc > 9) {
g_conf.LoadFeatureCost(argv[9]);
}
GBDT gbdt;
Elapsed elapsed;
gbdt.Fit(&d);
std::cout << "fit time: " << elapsed.Tell() << std::endl;
std::string model_file = train_file + ".model";
std::ofstream model_output(model_file.c_str());
model_output << gbdt.Save();
CleanDataVector(&d);
FreeVector(&d);
DataVector d2;
r = LoadDataFromFile(test_file, &d2);
assert(r);
elapsed.Reset();
DataVector::iterator iter = d2.begin();
PredictVector predict;
for ( ; iter != d2.end(); ++iter) {
ValueType p = Logit(gbdt.Predict(**iter));
predict.push_back(p);
}
std::cout << "predict time: " << elapsed.Tell() << std::endl;
std::string predict_file = test_file + ".predict";
std::ofstream predict_output(predict_file.c_str());
Auc auc;
for (size_t i = 0; i < d2.size(); ++i) {
predict_output << predict[i] << " " << d2[i]->ToString() << std::endl;
auc.Add(predict[i], d2[i]->label);
}
std::cout << "auc: " << auc.CalculateAuc() << std::endl;
auc.PrintConfusionTable();
CleanDataVector(&d2);
return 0;
}
void GBDT::Fit(DataVector *d) {
delete[] trees;
trees = new RegressionTree[g_conf.iterations];
size_t samples = d->size();
if (g_conf.data_sample_ratio < 1) {
samples = static_cast<size_t>(d->size() * g_conf.data_sample_ratio);
}
Init(*d, d->size());
for (size_t i = 0; i < g_conf.iterations; ++i) {
std::cout << "iteration: " << i << std::endl;
if (samples < d->size()) {
#ifndef USE_OPENMP
std::random_shuffle(d->begin(), d->end());
#else
__gnu_parallel::random_shuffle(d->begin(), d->end());
#endif
}
if (g_conf.loss == SQUARED_ERROR) {
for (size_t j = 0; j < samples; ++j) {
ValueType p = Predict(*(*d)[j], i);
(*d)[j]->target = (*d)[j]->label - p;
}
if (g_conf.debug) {
double s = 0;
double c = 0;
DataVector::iterator iter = d->begin();
for ( ; iter != d->end(); ++iter) {
ValueType p = Predict(**iter, i);
s += Squared((*iter)->label - p) * (*iter)->weight;
c += (*iter)->weight;
}
std::cout << "rmse: " << std::sqrt(s / c) << std::endl;
}
} else if (g_conf.loss == LOG_LIKELIHOOD) {
for (size_t j = 0; j < samples; ++j) {
ValueType p = Predict(*(*d)[j], i);
(*d)[j]->target =
static_cast<ValueType>(LogitLossGradient((*d)[j]->label, p));
}
if (g_conf.debug) {
Auc auc;
DataVector::iterator iter = d->begin();
for ( ; iter != d->end(); ++iter) {
ValueType p = Logit(Predict(**iter, i));
auc.Add(p, (*iter)->label);
}
std::cout << "auc: " << auc.CalculateAuc() << std::endl;
}
}
trees[i].Fit(d, samples);
}
// Calculate gain
delete[] gain;
gain = new double[g_conf.number_of_feature];
for (size_t i = 0; i < g_conf.number_of_feature; ++i) {
gain[i] = 0.0;
}
for (size_t j = 0; j < iterations; ++j) {
double *g = trees[j].GetGain();
for (size_t i = 0; i < g_conf.number_of_feature; ++i) {
gain[i] += g[i];
}
}
}