/*
* Create file with the position of each line
*/
void DataSet::create_position_file(const string& file) {
cout << endl;
cout << "Trying to create file with all line positions ..." << endl;
int pos = file.find(".");
string file_tmp = file.substr(0, pos);
string x_filename = file_tmp + ".pos_data";
string y_filename = file_tmp + ".pos_labels";
ofstream x_num_file(x_filename.c_str(), ios::binary);
ofstream y_num_file(y_filename.c_str(), ios::binary);
/* Try to open files */
ifstream xfp(x_filename_.c_str(), ios::binary);
if (!xfp) {
cout << "Could not open input file " << x_filename_ << endl;
exit(EXIT_FAILURE);
}
ifstream yfp(y_filename_.c_str(), ios::binary);
if (!yfp) {
cout << "Could not open input file " << y_filename_ << endl;
exit(EXIT_FAILURE);
}
/* Reading the header (first line of file)*/
int tmp;
xfp >> num_samples_;
xfp >> feature_dim_;
yfp >> tmp;
if (tmp != num_samples_) {
cout << "Number of samples in data and labels file is different" << endl;
exit(EXIT_FAILURE);
}
yfp >> tmp;
x_num_file << xfp.tellg();
x_num_file << "\n";
y_num_file << yfp.tellg();
y_num_file << "\n";
/* Going through complete files */
for (int n_samp = 0; n_samp < num_samples_; n_samp++) {
Sample sample;
sample.x = arma::fvec(feature_dim_);
yfp >> sample.y;
y_num_file << yfp.tellg();
y_num_file << "\n";
for (int n_feat = 0; n_feat < feature_dim_; n_feat++) {
xfp >> sample.x(n_feat);
}
x_num_file << xfp.tellg();
x_num_file << "\n";
}
xfp.close();
yfp.close();
x_num_file.close();
y_num_file.close();
}
/*
* Load complete dataset into memory
*/
void DataSet::load_complete_dataset(const string& x_filename,
const string& y_filename) {
/* Try to open files */
ifstream xfp(x_filename.c_str(), ios::binary);
if (!xfp) {
cout << "Could not open input file " << x_filename << endl;
exit(EXIT_FAILURE);
}
ifstream yfp(y_filename.c_str(), ios::binary);
if (!yfp) {
cout << "Could not open input file " << y_filename << endl;
exit(EXIT_FAILURE);
}
cout << endl;
cout << "Loading data file: " << x_filename << " ... " << endl;
cout << "Loading data file: " << y_filename << " ... " << endl;
/* Reading the header (first line of file)*/
int tmp;
long int tmp_samples;
xfp >> tmp_samples;
num_samples_ = tmp_samples;
xfp >> feature_dim_;
yfp >> tmp;
if (tmp != tmp_samples) {
cout << "Number of samples in data and labels file is different" << endl;
exit(EXIT_FAILURE);
}
yfp >> tmp;
/* Delete list with data points */
if (!add_points_)
samples_.clear();
set<int> labels;
/* Going through complete files */
for (int n_samp = 0; n_samp < num_samples_; n_samp++) {
Sample sample;
sample.x = arma::fvec(feature_dim_);
yfp >> sample.y;
labels.insert(sample.y);
for (int n_feat = 0; n_feat < feature_dim_; n_feat++) {
xfp >> sample.x(n_feat);
}
samples_.push_back(sample);
}
xfp.close();
yfp.close();
num_classes_ = labels.size();
if (random_) {
srand(init_seed());
random_shuffle(samples_.begin(), samples_.end());
}
}
void DataSet::load(const string& x_filename, const string& y_filename) {
ifstream xfp(x_filename.c_str(), ios::binary);
if (!xfp) {
cout << "Could not open input file " << x_filename << endl;
exit(EXIT_FAILURE);
}
ifstream yfp(y_filename.c_str(), ios::binary);
if (!yfp) {
cout << "Could not open input file " << y_filename << endl;
exit(EXIT_FAILURE);
}
cout << "Loading data file: " << x_filename << " ... " << endl;
// Reading the header
int tmp;
xfp >> m_numSamples;
xfp >> m_numFeatures;
yfp >> tmp;
if (tmp != m_numSamples) {
cout << "Number of samples in data and labels file is different" << endl;
exit(EXIT_FAILURE);
}
yfp >> tmp;
m_samples.clear();
set<int> labels;
for (int nSamp = 0; nSamp < m_numSamples; nSamp++) {
Sample sample;
sample.x = VectorXd(m_numFeatures);
sample.id = nSamp;
sample.w = 1.0;
yfp >> sample.y;
labels.insert(sample.y);
for (int nFeat = 0; nFeat < m_numFeatures; nFeat++) {
xfp >> sample.x(nFeat);
}
m_samples.push_back(sample); // push sample into dataset
}
xfp.close();
yfp.close();
m_numClasses = labels.size();
// Find the data range
findFeatRange();
cout << "Loaded " << m_numSamples << " samples with " << m_numFeatures;
cout << " features and " << m_numClasses << " classes." << endl;
}
