0bcbc6010cb581aec78aac51dbfda7d1e6a89621,data/generate_arabic_voice_dataset.py,,,#,4

Before Change

""" Load train set """
X = X_data[np.array(trainind)-1] //////Train data

max_nb_variables = -np.inf
min_nb_variables = np.inf

for i in range(X.shape[0]):
    var_count = X[i].shape[-1]

    if var_count > max_nb_variables:
        max_nb_variables = var_count

    if var_count < min_nb_variables:
        min_nb_variables = var_count

print("max number variables : ", max_nb_variables)
print("min number variables : ", min_nb_variables)

X_train = np.zeros((X.shape[0], X[0].shape[0], max_nb_variables))
y_train = y_data[(np.array(trainind)-1)] //////Train Labels

// pad ending with zeros to get numpy arrays
for i in range(X_train.shape[0]):
    var_count = X[i].shape[-1]
    X_train[i, :, :var_count] = X[i]

X_train_mean = X_train.mean()
X_train_std = X_train.std()
print("Train Mean +- std : ", X_train_mean, X_train_std)
//X_train = (X_train - X_train_mean) / (X_train_std + 1e-8)

""" Load test set """

X = X_data[(np.array(testind)-1)] ////////Test Data

X_test = np.zeros((X.shape[0], X[0].shape[0], max_nb_variables))
y_test = y_data[(np.array(testind)-1)] ////////Test Labels​

// pad ending with zeros to get numpy arrays
for i in range(X_test.shape[0]):
    var_count = X[i].shape[-1]
    X_test[i, :, :var_count] = X[i]

//X_test = (X_test - X_train_mean) / (X_train_std + 1e-8)

After Change

""" Load train set """
X = X_data[np.array(trainind)-1] //////Train data

var_list = []
for i in range(X.shape[0]):
    var_count = X[i].shape[-1]
    var_list.append(var_count)

var_list = np.array(var_list)
max_nb_variables = var_list.max()
min_nb_variables = var_list.min()
median_nb_variables = np.median(var_list)

print("max nb variables train : ", max_nb_variables)
print("min nb variables train : ", min_nb_variables)
print("median nb variables train : ", median_nb_variables)


X_train = np.zeros((X.shape[0], X[0].shape[0], max_nb_variables))
y_train = y_data[(np.array(trainind)-1)] //////Train Labels

// pad ending with zeros to get numpy arrays
for i in range(X_train.shape[0]):
    var_count = X[i].shape[-1]
    X_train[i, :, :var_count] = X[i]

X_train_mean = X_train.mean()
X_train_std = X_train.std()
print("Train Mean +- std : ", X_train_mean, X_train_std)
//X_train = (X_train - X_train_mean) / (X_train_std + 1e-8)

""" Load test set """

X = X_data[(np.array(testind)-1)] ////////Test Data

X_test = np.zeros((X.shape[0], X[0].shape[0], max_nb_variables))
y_test = y_data[(np.array(testind)-1)] ////////Test Labels​

// pad ending with zeros to get numpy arrays
for i in range(X_test.shape[0]):
    var_count = X[i].shape[-1]
    X_test[i, :, :var_count] = X[i]

//X_test = (X_test - X_train_mean) / (X_train_std + 1e-8)

Instances