I am trying to predict the sales of a product whose past 5 years data is available to me and some independent variable are also available (like the target market size, marketing budget, price, etc.). So now I can't use ARIMA or some time series model nor can I simply use regression to predict the sales. How should I go forward with this?
$\begingroup$
$\endgroup$
2
-
$\begingroup$ Are your independent variable data concurrent, or are you trying to forecast future sales based on current IV data? Do you want to forecast the IV data & then predict sales? Something else? $\endgroup$– gung - Reinstate MonicaCommented Jun 2, 2017 at 14:55
-
$\begingroup$ The IV are concurrent with the past sales data $\endgroup$– KaranCommented Jun 3, 2017 at 15:55
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
You need a model that will both use independent predictors AND take account of the autocorrelation issues.
A dynamic regression model will do this. There's a good explanation with an example (and snippets of R code) in Hyndman and Athanasopoulos's book, section 9.1: https://www.otexts.org/fpp/9/1
The basic steps are these (from the source cited): 
answered Jun 2, 2017 at 12:38