topic Linear (Holt) Exponential Smoothing in Discussions

Linear (Holt) Exponential Smoothing

bab02rs1 — Fri, 09 Jun 2023 00:33:20 GMT

My experience with Time Series in JMP is relatively short, so forgive me if this question seems basic.

Forecasting in the covid environment can be challenging at times due to lockdowns in various parts of the country. I’m trying to run the Linear (Holt) Exponential Smoothing for +2K locations based on most recent trends using Time Series – Autocorrelation by location. The results give me a time series by location but when I hold the shift +ctrl keys down, select simple smoothing, select linear exponential smoothing I have to set the predictive interval “0.95” and Constraints “Zero to One” for each location (+2K times).

Question: Is there a way to set this prompt once for all +2K locations?
Product Used: JMP (not JMP Pro)

Thank you

Re: Linear (Holt) Exponential Smoothing

ian_jmp — Thu, 06 May 2021 14:40:18 GMT

If the 'broadcast' modifier key isn't working in this case, you should be able to get what you want through some small changes to the script that the Time Series platform generates automatically. For example (and with only two levels of the 'By' variable in this case):

NamesDefaultToHere(1);

dt = Open("$SAMPLE_DATA/Big Class.jmp");
ts = dt << Time Series(
			Y( :height ),
			By( :sex ),
			// Add a line to specify the smoothing method and its parameters
			Simple Exponential Smoothing( Zero to One )
		);

Re: Linear (Holt) Exponential Smoothing

bab02rs1 — Thu, 06 May 2021 19:30:37 GMT

Ian, this was very helpful. Thank you!