1
$\begingroup$

I would like to forecast likely future demand based on historical demand. The problem is: I have no mathematical background and in relevant tutorials and even in literature formulas are used, that depend at every step on a given value for that time period, or in other words you need to know the value that you are forecasting, which seems either pointless or the word "forecasting" is used different in this context than in everyday language.

I am sure you know common models like Holt or Holt-Winters. For trended data, formulas are:

Current Forecast = alpha * Current demand + (1-alpha) * (Last Demand + Last trend)

Current trend = beta * (Current Forecast - Last Forecast) + (1-beta) * Last trend

As you can see, current trend depends on current forecast, and current forecast depends on current demand.

What do I do when I want to forecast future values for demand of a product when it has a trend? Please imagine the following list as a table because I dont know how to format the text properly:

Time period: 1, 2, 3, 4, 5, 6

Demand of waffles: 10, 11, 10, 13, 15, 16

I know I just could calculate a trend line in excel, but for several reasons I need to understand forecasting models (in a second step also for seasonal data).

$\endgroup$

closed as unclear what you're asking by Bob Jansen Oct 17 '15 at 13:05

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • $\begingroup$ Hi Gertrude Mauzikowski, welcome to Quant.SE! Without knowing what you want to forecast it will be hard to say, furthermore this feels quite basic. $\endgroup$ – Bob Jansen Oct 17 '15 at 13:05
  • $\begingroup$ Yes ..and linking to youtube videos feels a bit inefficient ... $\endgroup$ – Richard Oct 17 '15 at 14:04
  • $\begingroup$ Thank you both for your fast responses. I changed my question according to your comments. I can imagine this could be a rather basic problem, but it is difficult to get an intitial understanding of that topic, I guess because the word "forecasting" does not mean "predicting the future" here, but instead something like "adjusting parameters of my model that give the best fit to historical data". Anyway, in my job I need to predict future values. I think I have to linearly extrapolate using trend and level of the last time period, but this is NOWHERE stated clearly in forecasting literature. :/ $\endgroup$ – Gertrude Mauzikowski Oct 17 '15 at 17:45
  • $\begingroup$ With some rephrasing I think this fits better on stats. Please read their faq to conform with the requirements on a question there. $\endgroup$ – Bob Jansen Oct 18 '15 at 7:47
  • $\begingroup$ Note: Migrated to Cross Validated by the user. $\endgroup$ – Bob Jansen Oct 18 '15 at 13:05

Browse other questions tagged or ask your own question.