void SplitterInputReader::MoveToNextTrainSession() { while (in.HasNextSession()) { ++num_sessions_total; Session* s = in.GetNextSession(); if (!valid_users.count(s->user_id)) { ++num_sessions_skipped; // Invalid user. Skipping. continue; } int day_id = s->day_id ; assert(day_id > 0 && day_id <= DAYS_UNDER_OBSERVATION); if (is_in_train[day_id - 1]) { next_session = s; return; } else { test_data_writer.Print(*s); if (s->HasSwitches()) { expected_sessions_with_switches.push_back(s->session_id); } delete s; } } next_session = NULL; // End of sessions test_data_writer.Close(); }
void SplitterInputReader::PrepareMetadata(const MarkedData &data) const { //map<int, set<int> > users_to_sessions; // maps user_id to set of all days where any sessions are found map<int, set<int> > users_to_sessions_with_switches; // same but only sessions with switches are counted set<int> user_ids; PlainInputReader temp_input(data); while (temp_input.HasNextSession()) { Session* s = temp_input.GetNextSession(); user_ids.insert(s->user_id); // users_to_sessions[s->user_id].insert(s->day_id); if (s->HasSwitches()) { users_to_sessions_with_switches[s->user_id].insert(s->day_id); } delete s; } FILE* out_file = fopen(GetMetadataFileName(data).c_str(), "wb"); fprintf(out_file, "%d\n", (int)user_ids.size()); for (set<int>::iterator it = user_ids.begin(); it != user_ids.end(); ++it) { int user_id = *it; fprintf(out_file, "%d", user_id); set<int> days; /* days = users_to_sessions[user_id]; fprintf(out_file, " %d", (int)days.size()); for (set<int>::iterator it2 = days.begin(); it2 != days.end(); ++it2) { fprintf(out_file, " %d", *it2); } */ days = users_to_sessions_with_switches[user_id]; fprintf(out_file, " %d", (int)days.size()); for (set<int>::iterator it2 = days.begin(); it2 != days.end(); ++it2) { fprintf(out_file, " %d", *it2); } fprintf(out_file, "\n"); } fclose(out_file); }