From theory to practice - page 352

New comment
 
Alexander_K2:

Mm-hmm. Thank you. Now that's a conversation of business.

ZZZ all of this, by the way, applies not only to NS, but to any other prediction method.)

 
Novaja:

if we try to logarithm these increments and construct a histogram of the distribution, will it approach a Gaussian? This is also an option.

This discovery, if I am not mistaken, was made about 10-15 years ago in a paper at a RTS conference.

 
Yuriy Asaulenko:

In the general case, never. However, prediction is possible in some limited areas. If you can make the NS work only in those areas, then prediction is possible.

If all the evil lies in non-stationarity, so why not try to at least somehow restabilize the process? After all, everyone is talking about it, what is it, everything is lame, there is no point of reference, even a sliding window to cling to, what is there to cling to?

 
Yuriy Asaulenko:

This discovery, if I am not mistaken, was made about 10-15 years ago in a report at a RTS conference.

So great, if we have a good exponent, let's reverse it, what's the point?

 
Novaja:

If all the evil is in the non-stationarity, why not try to make it stop somehow? After all, everyone is talking about it, what is it, everything is lame, there is no point of reference, even a sliding window to cling to, what is there to cling to?

And why make it stationary at all? Stationarity by itself won't help you at all for any prediction.

To be clearer, random walks (at the level of increments) are stationary processes. So what's the deal with forecasting?

For forecasting, apart from stationarity, other requirements must be met. Why the hell would anyone think that removing non-stationarity these requirements "by magic" will be fulfilled. They won't. It cannot be because it can never be.

 
Novaja:

So fine, if we have a good exponent, let's reverse it, what's the point?

So do it.)

 
Yuriy Asaulenko:

Just to be clear, random walks are a stationary process. And what's the prediction?

stationary - oscillating around the mean value. sb is not stationary.
http://sernam.ru/book_tp.php?id=95

 
igrok333:

stationary - oscillating around the mean value. sb is not stationary.
http://sernam.ru/book_tp.php?id=95

Is it? What about the generative process? Nevertheless, corrected my post.

 
Novaja:

So fine, if we have a good exponent, let's reverse it, what's the point?

To reverse the noise distribution back to BP even Master of Physical Sciences Alexander does not dream.))) Or maybe he does dream, but he won't admit it))) That's at least a Nobel Prize, and maybe even a gold medal for the best VR visionary.))
 
Andrei:
Even master of physical sciences Alexander does not dream of turning noise distribution back to BP.))) Or may be he does dream, but he won't admit it))) After all it is at least a Nobel, and maybe even a gold medal for the best VR visionary.))

That's the thing, he does, everyone has been talking about it for the last nn pages of this forum.

1...345346347348349350351352353354355356357358359...1981
New comment