5

I am working on a project, and I am absolutely new to forecasting and not so strong in statistics. I have an employee data for the last 7 years, along with the other variables like economic growth, employee turnover, vacancies, and some other economical factors. I have to do forecasting for the next 5 years. I have some questions:

  • Is it possible to do a time series analysis with more than one explanatory variable? Can this be done in R?

I would appreciate any kind of help. Thanks in advance!!

S.B
  • 151
  • 1
  • 3
  • 8
  • See [Hyndman & Athanasopoulos, *Forecasting: principles and practice* Ch. 9.1](https://www.otexts.org/fpp/9/1) for an introduction to time-series regression with examples using R. – Scortchi - Reinstate Monica Apr 25 '16 at 10:21
  • Can this question be reopend again please? I really need some help with the time series models in combinations with explanatory variables. – S.B Apr 25 '16 at 10:31
  • @Scortchi, perhaps *off-topic* is not a very good characterization. IMHO, the first question is valid but too basic (there is no such reason for closing, though), the second one is *off topic* while the third one is *too broad*. What if the OP deleted the last two questions? Would that be a good enough ground for reopening? – Richard Hardy Apr 25 '16 at 10:39
  • S.B., yes, time series modelling is possible with multiple explanatory variables. Multiple regression (possibly with ARMA errors), autoregression (possibly with exogenous variables) and vector autoregression (possibly with exogenous variables) could be your starting points. – Richard Hardy Apr 25 '16 at 10:41
  • @RichardHardy: The original version, prior to edits, contained only the 2nd question. I've re-opened the question, though it's perhaps rather broad even when focused on regression: a good answer should IMO include helpful references for someone new to time series analysis. – Scortchi - Reinstate Monica Apr 25 '16 at 10:44
  • It's *very clear* that time series analysis with multiple explanatory variables *is* possible. I suggest you focus more on *how* it can be done best *in your instance*. – Richard Hardy Apr 25 '16 at 11:56
  • Thanks for the answers. Yes it is clear now. Do you mean that I should change the question in how it can be done? Do you recommend any books or literature that focuses on multiple time series regression? I'm having problems with choosing a model in R and how to apply the model in my problem.. Also.. I don't have many observations. How do you know how many observations you need for a good time series analysis? – S.B Apr 25 '16 at 12:03

0 Answers0