0
$\begingroup$

I have a dataset containing values representing monthly sales in hypothetical e-commerce website. First columns contains order of particular month (eplanatory variable) and the second column (dependent variable) is value of sales in this month.

month_order value
    1   2591
    2   2262
    3   2531
    4   2379
    5   2106
    6   2500
    7   2345
    8   2200
    9   2385
    10  2991
    11  2219
    12  2600
    13  3000

Equation of linear regression for this data set is:

f(x) = 27,42x + 2278

Question: Because I would like to make a very simple prediction of sales value in next month and I have no another more suitable eplanatory variable, may I use month_order and predict linear trend? In my case it would be:

f(14) = 27,42 * 15 + 2278 = 2689

is this correct in a theory of a linear regression and can I count it in this way?

| cite | improve this question | |
$\endgroup$
1
$\begingroup$

The only issue here is an assumption violation; i.e. correlated errors. You would probably be better served using autoregressive or moving-average methods rather than a simple linear extrapolation.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Could you be more specific what particular method I can use? I am not depended on linear regression only, I can use another method if is more suitable. $\endgroup$ – Artegon Mar 18 '13 at 14:57
  • $\begingroup$ There are several steps in determining the appropriate model. I would strongly suggest googling "time series analysis" in order to see what is involved. $\endgroup$ – ReliableResearch Mar 18 '13 at 16:27

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.