在python3.X中，pandas已经没有了MovingOLS。因此，若想求解滚动窗口的OLS，只能自己来写。

原来的方法如下：

model = pd.stats.ols.MovingOLS(y=df.y, x=df.x, window_type='rolling', window=1000, intercept=True)

替换的代码如下：

window = 1000


a = np.array([np.nan] * len(df))


b = [np.nan] * len(df)  # If betas required.

 y_ = df.y.values


x_ = df[['x']].assign(constant=1).values


for n in range(window, len(df)):
    

      y = y_[(n - window):n]
    

      X = x_[(n - window):n]
    

      # betas = Inverse(X'.X).X'.y
    

      betas = np.linalg.inv(X.T.dot(X)).dot(X.T).dot(y)


      y_hat = betas.dot(x_[n, :])
    

     a[n] = y_hat

     b[n] = betas.tolist()  # If betas required.





或者：


df=df.dropna() #uncomment this line to drop nans

window = 5

df['a']=None #constant


df['b1']=None #beta1


df['b2']=None #beta2


for i in range(window,len(df)):
    

    temp=df.iloc[i-window:i,:]
    

     RollOLS=sm.OLS(temp.loc[:,'Y'],sm.add_constant(temp.loc[:,['time','X']])).fit()     
    

    df.iloc[i,df.columns.get_loc('a')]=RollOLS.params[0] 
    

    df.iloc[i,df.columns.get_loc('b1')]=RollOLS.params[1]
    

    df.iloc[i,df.columns.get_loc('b2')]=RollOLS.params[2]





当然，也有人自己写了一个模型解决这个问题，如下：




# Rolling regressions



from pyfinance.ols import OLS, RollingOLS, PandasRollingOLS


y = data.usd


x = data.drop('usd', axis=1)



window = 12  # months


model = PandasRollingOLS(y=y, x=x, window=window)

print(model.beta.head())





参考：主要是stackoverflow里面的2个网址

/questions/44380068/pandas-rolling-regression-alternatives-to-looping

/questions/44707384/python-pandas-has-no-attribute-ols-error-rolling-ols
https://www.e-learn.cn/content/wangluowenzhang/754972

https://e-learn.cn/content/wangluowenzhang/192368