Econometrics: one step ahead forecast - page 43

New comment
 
avtomat:


I remember you promising a model in state space on Monday.
 
Avals:

It was not a question of choosing the sample size to analyse, but of the forecast horizon. I don't think it should be fixed in time, but if you really want to discuss what its value depends on, then the size of the sample to be analysed is one of the factors
Yes you are right. It's quite enough to be one step ahead.
 
faa1947:
I remember you promised me a model in the state space on Monday.

First of all, I don't owe you anything. Secondly, I have other pressing concerns. Thirdly, my statement included the proviso "if nothing interferes". Fourthly, your attitude to techies is not conducive to an atmosphere of mutual assistance. And fifthly, considering your level of knowledge you should thoroughly study such model, and for this you need a lot of time.

In particular, familiarity with the theory of differential equations would be required -- how well do you master this matter?

zy.

And I'll make an example of model in state space - it will be useful. Now I'm doing another experiment -- I'll use this real example of actual data to build a model. But it's a slow process -- I'm going to fit the model to the process, not the process to the model.

 
avtomat: And I will make an example of the model in the state space - it will be useful. Now I'm doing another experiment - I'm going to use this real example of actual data to build a model. But it's a slow process -- I'm going to fit the model to the process, not the process to the model.

Looking forward to
 

And I used to think it was stupidly commutative all the time:

Подгонять( модель, процесс ) = Подгонять( процесс, модель ).

Never mind, keep discussing...

 
Mathemat:

And I used to think it was stupidly commutative all the time:

Подгонять( модель, процесс ) = Подгонять( процесс, модель ).

Never mind, keep discussing...

Absolutely not!
 
It's such a good way to start the day!
 
avtomat: Absolutely not!
And what do you yourself see as the fundamental differences between these... er... functions? I, for example, am still frankly confused.
 
Mathemat:
And what do you yourself see as the fundamental differences between these... er... functions? I, for example, am still frankly confused.

Simply put, the process is primary, the model is secondary. The reverse is not true.
 
Mathemat:

And I used to get the impression that everything here is stupidly commutative:

Подгонять( модель, процесс ) = Подгонять( процесс, модель ).

Never mind, keep discussing...

IMHO


Empirical approach:

look at the data

come up with a model that can describe the observed phenomena

estimate the parameters of the model using the available data

if the model works poorly, we devise another model

Theoretical approach:

build a theory which presumably results in the observed process

build a model for optimal trade (prediction) on the basis of the proposed theory

estimate the parameters of the model using the data available

If the model works poorly - think out another theory

1...363738394041424344454647484950...139
New comment