如何训练多重回归 Model 并在 Python 中获取估计结果？

Question

I have a dataframe. This dataset contains data of my company and knowledge of my competitors. It is look like:

Date         a_mine   b_mine  a_comp  b_comp  c_mine  c_comp   
1.01.2020    17.328   6.736   10.592  66.836  3.15    3.15
1.02.2020    16.680   6.522   10.158  64.097  3.46    3.45
1.03.2020    13.616   5.334   8.282   58.554  3.76    3.75
1.04.2020    8.351    3.075   5.276   37.301  3.76    3.75
1.05.2020    13.610   5.837   7.773   54.955  3.76    3.76
1.06.2020    14.361   5.875   8.486   59.996  3.79    3.80


a_mine: Net sales of my company
a_comp: Net sales of competitors
b_mine: bonus sales of my company
b_comp: bonus sales of competitors
c_mine: unit price of my product
c_comp: unit price of competitors product

I want to find bonus sales effect on the net sales and finally, I want to create a result table like this (an example results):

Component  Parameter  Estimate  Standart_error  t_value  Approx Pr>|t|
a_mine     constant   485052.1  22517.1         21.58    < 0001
b_mine     scale      1.15365   0.12745         9.07     < 0001

I tried to train my model with multiple linear regression. But I could not success for this.

How to train my model and get this results in python?

Answer 1

You can always use statsmodels ' OLS regression , which has a .summary() method that returns you the table you need:

Y = df.a_mine
X = df[["b_mine", "c_mine"]]

model = sm.OLS(Y, X)
results = model.fit()

results.summary()