def divide_save(filename, savename, flag_pca=False, portion=0.8,
permu_flag=True, usage_ratio=1, n_labels=12, n_seq=20):
X_, y_ = csv_read(filename)
if flag_pca:
pca = PCA(n_components=8, whiten=True)
X_ = pca.fit_transform(X_)
X, y = csv_reformat(X_, y_, filename, permu_flag=permu_flag, n_labels=n_labels, n_seq=n_seq)
print X.shape, y.shape
n_samples, n_dimension = X.shape
n_use_samples = int(ceil(n_samples * usage_ratio))
train_part = int(ceil(n_use_samples * portion))
X_train = X[:train_part]
X_test = X[train_part:n_use_samples]
y_train = y[:train_part]
y_test = y[train_part:n_use_samples]
train = np.concatenate((X_train, y_train), axis=1)
test = np.concatenate((X_test, y_test), axis=1)
train_name = path + 'train_' + savename + '.csv'
test_name = path + 'test_' + savename + '.csv'
print 'Saving divided data..'
np.savetxt(train_name, train, delimiter=',')
np.savetxt(test_name, test, delimiter=',')
def divide_data(portion=0.8, permu_flag=True, usage_ratio=1, n_seq=20):
X, y = csv_reformat(permu_flag=True, n_seq=n_seq)
n_samples, n_dimension = X.shape
n_use_samples = int(ceil(n_samples * usage_ratio))
train_part = int(ceil(n_use_samples * portion))
X_train = X[:train_part, :]
X_test = X[train_part:n_use_samples, :]
y_train = y[:train_part]
y_test = y[train_part:n_use_samples]
X_train = np.reshape(X_train, (-1, X_train.shape[1]/n_seq, n_seq))
X_test = np.reshape(X_test, (-1, X_test.shape[1]/n_seq, n_seq))
return X_train, y_train, X_test, y_test