class Eystd(CustomFactor):  
    # Pre-declare inputs and window_length  
    inputs = [morningstar.valuation_ratios.earning_yield]  
    window_length = 1260   # past five years  
    def compute(self, today, assets, out, ey):  
        eystd = np.std(ey, axis=0)  
        #eystd[eystd == 0] = 1  
        out[:] = eystd  

the code sample above simply calculates the past five years standard deviation of fundamental data earning_yield in pipeline for every stock in the universe.
Real earning_yield = earning_yield - USGG 10y treasury yield is the indicator I want and since Quantopian doesn't have US treasury yield data, I use fetch_csv to get it from quandl:
fetch_csv('https://www.quandl.com/api/v3/datasets/USTREASURY/YIELD.csv?api_key=MMrssLMT_-ToHwUcxf_m', date_column='Date', symbol='10_Year', pre_func = preview, post_func=rename_col, date_format='%Y-%m-%d') It works very well when I just need current 10y_treasury_yield data individually.
current_yield = data.current('10_Year','10_Year_Yield') log.info(current_yield)

My problem is how can I use it as a input data source in pipeline calculation for example: inputs = [morningstar.valuation_ratios.earning_yield] ** and we can use **inputs = [data.history('10_Year','10_Year_Yield')]. So that I can calculate the standard deviation of Real earning_yield = earning_yield - USGG 10y treasury yield and get the output in pipeline.

Is there anyone know the solution? Great thanks!