Linear regression channel - page 14

New comment
 

"Santa Barbara" something...

(brightens up my evening!)

 
Yuriy Asaulenko:
As if anyone but Dimitri had any doubts.
That's just it, fxsaber had doubts.
 
Nikolai Semko:
I'm jealous. Fruitful work!
My message was that you can do this not only with RMS for a simple waving machine, but also for a polynomial of any degree.
You can't do it for a polynomial. The regression line is almost completely rearranged at every price change.
 
Yuriy Asaulenko:
For a polynomial you will not. The regression line is almost completely rebuilt at every price change.
already did 4 years ago for any degree and put it on the market.
 
Nikolai Semko:
already did it four years ago and put it on the market.
So it's not a polynomial regression line.
And if not, there's no question about it on a polynomial.
 
Yuriy Asaulenko:
You're welcome. I wasn't going to race you. I just asked them to stay out of the way, and for those in the know, to be quiet.
The algorithm, yes, is almost equivalent, I wrote about it on page 1-2 of the thread.
Dimitri, sorry, although proven, I can't accept Hennessy, to my regret.
The code, I suppose, makes no sense to write. Come Semko, and that's it... Priority wanted)). As if anyone but Dimitri had any doubts.
I will have to go to the shop myself.

What's the priority? This calculation is a hundred years old,
If anyone remembers the Algorithms and Programs Foundation (Minsk), it was published in one of the first issues (and they seem to have been only for the EC).
EC is IBM 360/370 - for those who are young)

 
Mikhail Dovbakh:

What's the priority? This calculation is a hundred years old,

No doubt about it. But to get into the algorithm pool for that. I'm freaking out.)
The intelligence of gentlemen scientists is sometimes frightening.
 
Yuriy Asaulenko:
No doubt about it. But, to get into the algorithm fund for that. I'm spoiled. ))

Nobody climbed) From memory, the link to the priority.
And the link to the blog I mentioned, Google kindly gave immediately to "Single-pass calculation of the RMS" ...

)

 
Nikolai Semko:
That's the thing, fxsaber doubted.

That's right. For what reason the fifth-grade problem(revealing the square of the difference) is not seen, I do not understand.

There's just one thing I don't understand. Why couldn't you say the highlighted one on the first page of the discussion? It is, to put it mildly, ugly.

 
fxsaber:

That's right. For what reason the fifth-grade problem(revealing the square of the difference) is not seen, I do not understand.

There's just one thing I don't understand. Why couldn't you say the highlighted one on the first page of the discussion? It is, to put it mildly, ugly.

In other words, I have committed an abomination by suggesting the possibility of increasing the speed of the algorithm by orders of magnitude, but not providing the code with a ready-made solution?
You know, I'm studying to be a programmer in Ottawa. And when my classmates ask me before the deadline to submit the code, of course I submit it, but every time I feel like I'm ripping them off by giving them a ready-made solution.

And by the way, the discussion on the first page wasn't about squaring the difference. The RMS of linear regression is a bit more complicated than the RMS of a conventional waving machine. And, as you can see, the RMS of the regular oscillator (Bollinger Bands to be exact) is not justthe difference of squares of extreme values.
The fact that I have made some people think, having created intrigue, is more useful than posting a ready-made solution. I already know from experience that most people pass by ready-made solutions. You of all people should know that.

1...789101112131415161718
New comment