def apply_transformation(dataframe, train_target):
    """
    Apply transformation on features and calculate the correlation with the target
    :param dataframe: pandas dataframe
    :param train_target: pandas series
    :return: pandas dataframe contains the correlation between each feature and the target for different
     applied transformations
    """
    # remove negative values and zeros to avoid math problem
    for col_i in dataframe.columns:
        dataframe[col_i] += abs(min(dataframe[col_i])) + 1

    # 1 means the original values. If the type is number, it means x^number
    transformation_type = [1, "log", 0.25, 0.5, 0.75, 2, 3, 4]
    correlation_dataframe = pd.DataFrame(columns=[str(x) for x in transformation_type])

    for trans_i in transformation_type:
        if trans_i == "log":
            dataframe_trans = np.log(dataframe)
        else:
            dataframe_trans = (dataframe) ** trans_i

        correlation_dataframe[str(trans_i)] = [round(np.corrcoef([dataframe_trans[x], train_target])[0][1], 2)
                                          for x in dataframe.columns]
    return correlation_dataframe