Input contains NaN, infinity or a value too large for dtype('float64').

regr = linear_model.LinearRegression()
regr.fit(df_input[['a','b','c','d','e','f','g']], df_input['h'])

线性拟合的时候已经查过了数据没有null，但是还报错说
Input contains NaN, infinity or a value too large for dtype('float64').

打印一下你的df_input我看看