0

I only have 2 available points and I wish to forecast its value for the next year. I'm currently looking into exponential smoothing, and econometric input-output model. Which one would be more appropriate? This is about poverty.

Richard Hardy
  • 54,375
  • 10
  • 95
  • 219
Katherine
  • 29
  • 1
  • 2
    With just two points you don't have all these choices. All you have is an estimated slope which means that prediction can only be done by extending the straight line. When you get more data you may have other options. – Michael R. Chernick Mar 28 '17 at 13:40
  • @Michael Chermick Can I possibly use exponential smoothing since what I only need there is an actual and forecasted value? – Katherine Mar 28 '17 at 13:43
  • 2
    Another option somewhat less intrusive in my opinion is to simply average the values. This seems to me to be between fitting a trend line and using the last value. – IrishStat Mar 28 '17 at 13:43
  • ...although the forecast itself with not be between the forecast from the trend and the last value (just to note). – Richard Hardy Mar 28 '17 at 13:47
  • 1
    I think all the mentioned ideas are plausible but it is unclear which will be most accurate. – Michael R. Chernick Mar 28 '17 at 13:47
  • Please clarify what you mean by "I only have 2 available points." – rolando2 Mar 28 '17 at 14:31
  • I'm seeing that you asked essentially the same question 3 hours earlier at http://stats.stackexchange.com/questions/270287/causal-forecasting - please put all your relevant info into one and delete the other. – rolando2 Mar 28 '17 at 14:47

0 Answers0