mhaze
Banned
- Joined
- Jan 10, 2007
- Messages
- 15,718
Putting ForecastingPrinciples.COM to the Test?
Agreed. Where it gets muddy is whether (1) the modelers do a decent job of forecasting climate based on the applicable subset of principles fronted by Armstrong (2) whether Chapter 8 does a decent job of summarizing the predictive merits of these models and their faults (again based on the applicable subset of principles).
econometric methods are not confined to economic problems, but can be applied to forecasting in situations where you have theories about what causes changes in the thing that you are forecasting. In order to forecast using econometrics, you will also need to have data on the causal variables. These data need to vary, and to make the exercise worthwhile you will need to be able to forecast the causal variables sufficiently accurately that the overall error attained is smaller than could be achieved by extrapolation.
How do we understand the merits (or lack of) of Armstrong's method? I come up with three primary sources, and a method.
The sources are:
And the Method is:
The "Forecasting Audit" is the method used to evaluate the target (Chapter 8, IPCC here) I set up a login and password to do an audit. NO GO with Firefox, IE worked like a champ.
That's done here. As used by Armstrong, a "Forecasting Audit" is the auditor answering a set of questions in more than a dozen categories based on the "formal procedure". It's a simple issue of whether the formal procedure follow the standard?
Questions are answered "Not Applicable", -2 to +2 for "following the standard", or "question mark". After completing the audit, the audit is saved, and the user can later log on again and edit the work. Perhaps the first time around, some sections were not understood, and additional review of the material was required prior to completing them. These can be answered "?". "?" also brings up an explanatory page on that forecasting principle.
Next do a test "Forecasting Audit" - and this answers several of your questions right off the bat.
Principles can be marked as "Not applicable".
Double click to enlarge.
Agreed. Where it gets muddy is whether (1) the modelers do a decent job of forecasting climate based on the applicable subset of principles fronted by Armstrong (2) whether Chapter 8 does a decent job of summarizing the predictive merits of these models and their faults (again based on the applicable subset of principles).
Armstrong says -
econometric methods are not confined to economic problems, but can be applied to forecasting in situations where you have theories about what causes changes in the thing that you are forecasting. In order to forecast using econometrics, you will also need to have data on the causal variables. These data need to vary, and to make the exercise worthwhile you will need to be able to forecast the causal variables sufficiently accurately that the overall error attained is smaller than could be achieved by extrapolation.
How do we understand the merits (or lack of) of Armstrong's method? I come up with three primary sources, and a method.
The sources are:
- the FAQ.
- Armstrong's use of phrases-Forecasting Dictionary.
- The forecasting principles are here, divided into groups.
And the Method is:
The "Forecasting Audit" is the method used to evaluate the target (Chapter 8, IPCC here) I set up a login and password to do an audit. NO GO with Firefox, IE worked like a champ.
That's done here. As used by Armstrong, a "Forecasting Audit" is the auditor answering a set of questions in more than a dozen categories based on the "formal procedure". It's a simple issue of whether the formal procedure follow the standard?
Questions are answered "Not Applicable", -2 to +2 for "following the standard", or "question mark". After completing the audit, the audit is saved, and the user can later log on again and edit the work. Perhaps the first time around, some sections were not understood, and additional review of the material was required prior to completing them. These can be answered "?". "?" also brings up an explanatory page on that forecasting principle.
Next do a test "Forecasting Audit" - and this answers several of your questions right off the bat.
Principles can be marked as "Not applicable".
irrelevant (unlikely to apply when forecasting any physical system)
unhelpful (unlikely to be of use in the particular system under consideration)
unprincipled (eg attacks properties of the question, not the problem soln)
naive (wonderful in principle, but displaying an ivory tower separation from the facts on the ground)
agreed (already widely implemented)
on target (under-appreciated or overlooked, and carrying nontrivial implications for climate science)
Establish a login, then takes 30 seconds and is the route to understanding the "foreasting audit software". Following are the first several screens (pay NO attention to my answers for now, I was just plugging values in to see where this would all go to) -unhelpful (unlikely to be of use in the particular system under consideration)
unprincipled (eg attacks properties of the question, not the problem soln)
naive (wonderful in principle, but displaying an ivory tower separation from the facts on the ground)
agreed (already widely implemented)
on target (under-appreciated or overlooked, and carrying nontrivial implications for climate science)
Double click to enlarge.
Last edited:
.
. The Bali Conference is about talking about action. The conference that is about acting will take place on higher ground. Geneva springs to mind, but I'm a traditionalist.
.