utils.py 文件源码

def Orthogonalize(left_data, right_data, left_name, right_name):
    """?????
       ?????????[date,IDs,factor1,factor2...]
    ??
    --------
    left_name: str
        ??1?????
    right_name: str or list
        ??2??????????????????
    industry: str
        ??????????None?????????
    """

    def OLS(data, left_name):
        tempData = data.copy()
        yData = np.array(tempData.pop(left_name))
        xData = np.array(tempData)
        NaNInd = pd.notnull(yData) & pd.notnull(xData).all(axis=1)
        model = sm.OLS(yData, xData, missing='drop')
        res = model.fit()
        data[left_name+'_orthogonalized'] = np.nan
        data.ix[NaNInd, left_name+'_orthogonalized'] = res.resid
        return data

    factor_1 = left_data[[left_name]]
    if not isinstance(right_name, list):
        right_name = [right_name]
    factor_2 = right_data[right_name]
    factor = pd.concat([factor_1, factor_2], axis=1)
    factor['alpha'] = 1  # ?????
    factor = factor.groupby(level=0).apply(OLS, left_name=left_name)
    return factor[[left_name+'_orthogonalized']]