def normalization(x_train):
    '''
        apply normalization to the samples (x_train)
        returns a new array that has been properly normalized,
        also returns the mean and standard deviation per feature in the training set
    '''
    # >>> YOUR CODE STARTS HERE <<<
    x_train_normalized = np.empty(shape = x_train.shape)
    i = 0
    for features in x_train:
        normed = []
        feature_std = np.std(features, axis=0)
        feature_mean = np.mean(features, axis=0)
        for feature in features:
            normed.append((feature-feature_mean)/feature_std)
        x_train_normalized[i] = normed
        i+=1

    print(x_train_normalized)

    # array of normalized features
    mean = np.mean(x_train_normalized, axis=0) # array of mean value per feature
    std = np.std(x_train_normalized, axis=0) # array of std value per feature

    # >>> YOUR CODE ENDS HERE <<<

    return x_train_normalized, mean, std