bool AdClickPredictor::train()
{
// Obtain all the data files under the dataPath_
DIR *dp;
struct dirent *dirp;
if ( (dp = opendir(dataPath_.c_str())) == NULL )
{
LOG(ERROR) << "opendir error: " << dataPath_ << std::endl;
return false;
}
while ( (dirp = readdir(dp)) != NULL )
{
if ( (strcmp(dirp->d_name, ".") == 0) || (strcmp(dirp->d_name, "..") == 0))
continue;
if ( dirp->d_type == DT_DIR )
continue;
std::string filename(dirp->d_name);
LOG(INFO) << "Train AdClickPredictor from file: " << dirp->d_name << std::endl;
//Training model with the current data file
trainFromFile(filename);
}
return true;
}
void Classifier::train(const std::vector<Document> &trainDocs) {
std::ofstream data(TRAIN_FILE_NAME);
int size = 0;
std::vector<std::vector<ClassifiedTriple>> docs;
for (const Document &d : trainDocs) {
std::vector<ClassifiedTriple> triples = d.getClassifiedTriples();
size += triples.size();
docs.push_back(triples);
}
data << size <<" " << NUM_INP <<" " << NUM_OUTP<<"\n";
for (auto triples : docs) {
writeClassifiedTriples(triples,data);
}
data.close();
trainFromFile(TRAIN_FILE_NAME);
}
