0
$\begingroup$

enter image description here

Hi All, Im very new to stats and I was wondering if someone could help me make sense of the regression formula at the top of this workbook? invoice price is on the x axis and discount is on my y. thanks!

$\endgroup$
  • 1
    $\begingroup$ Not even vaguely useful. Take logarithms or otherwise transform the data to get rid of the heteroscedasticity. $\endgroup$ – Carl Oct 4 '19 at 1:46
0
$\begingroup$

There almost no relationship between your variables as the r^2 is 0.02.

If you were to try and model the total discount, you would start at 17% (y = 0.17) then add 0.00000001 million to for every dollar that you move along the x axis (the invoice price).

If you wanted to use these variable, perhaps you could try and normalise the data (getting the log of the invoice price perhaps) and then try and model it.

This seems to explain it simply: https://statisticsbyjim.com/regression/interpret-coefficients-p-values-regression/

$\endgroup$
  • $\begingroup$ There are negative % values (whatever that means), so the log won't work for the Y-axis. Cube root is an option. $\endgroup$ – mkt - Reinstate Monica Oct 4 '19 at 8:29
0
$\begingroup$

The correct interpretation is that this model doesn't fit and another should be used.

Given that the x variable is price, I think taking log(price) makes sense, as logs of monetary variables often do.

But even that model may not fit that well, as you will probably still have some extreme outliers. I therefore suggest not using linear regression but using quantile regression or robust regression.

If you are really new to statistics, you should probably hire a consultant.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.