Machine learning in trading: theory, models, practice and algo-trading - page 328
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
PSSP
There are more promising ones: random forests, a variety of ada.
How can you seriously compare R to skylab? Some kind of redneck package, not in any rankings...
What do you think I'm doing?
The forests are there too.
You should not roll on SciLab.) Unlike R, it has its own tasks, and, of course, it is less common, but is nevertheless widely used in universities and research organizations - MTI, Boing, Bell, etc. Of course R and SciLab do not replace each other and do not compete - the subject areas are different, however.
What do you think I'm doing?
There are forests, too.
You should not roll on SciLab.) Unlike R, it has its own tasks, and, of course, it is less common, but is nevertheless widely used in universities and scientific organizations - Boing, Bell, etc. Of course, R and SciLab don't replace each other and don't compete - the subject areas are different, though.
Agitating to increase the population of R supporters.
Agitating to increase the population of R supporters.
That's right.) I, on the other hand, offer an alternative. I suppose it's equivalent.) Something better, something worse.
Say, SciLab will be more interesting in terms of computational math. Stat. methods are also quite good there, but of course not compare with R.
That's right.) I, on the other hand, offer an alternative. I suppose it's equivalent.) Something better, something worse.
Say, SciLab is more interesting in terms of computational math. Statistical methods are rather good there too, but certainly can't be compared with R.
This Brownian motion in search is not interesting. Is it really so hard to look through all the articles, which are on this site? If you are interested in the process of searching, it's different. Decide what problems you want to solve (regression/classification?). In my opinion regression has no prospects.
The R language has everything you need for trading both forex and stocks. It's a great MT/R combination. Just experiment and implement it. And you propose to go where there is none of that.
Can you give me an example of computational math?
Good luck
Why are you picking on networks? They don't work and that's it, it's just a vogue of the last century, probably the first machine learning package that was available.
There are more promising ones: random forests, various ada. And generally the caret shell package, which has a couple hundred packages, including grids, and you can do automatic selection between them.
what is the variety of hell? ) Until you learn it all you'll grow old and die without profit, it's real hell
what is the variety of hell? ) You'll grow old while you learn it all.
Don't listen to anyone. There is no proof that any scaffolding or anything else works better than networks.
But there is evidence that a network can approximate any function, but I have not seen such a proof for the same scaffolding.
If a network can't do it, the forests certainly can't. Moreover, as I see, your results are decent.
So saying "optimization is dangerous" is like saying "a microscope is dangerous" - it hurts to hit you on the head with it.
What is a variety of hell? ) You'll grow old while you learn it all.
Don't listen to anyone. There is no proof that any scaffolding or anything else works better than networks.
But there is evidence that a network can approximate any function, but I have not seen such evidence for the same scaffolding.
If a network can't do it, the forests certainly can't.
Actually, a random forest is a classification and doesn't do approximation at all
In my opinion, regression has no prospects.
AND GARCH?
In classification, everything rests on a set of predictors. It is not clear where to look.
And in GARCH it's a dumb process: you model a trend, analyze the residual - model it, analyze the aggregate model residual - model this residual - some process without excessive creativity and guesswork.
Don't listen to anyone. There is no proof that any scaffolding or anything else works better than networks.
But there is evidence that a network can approximate any function, but I have not seen such a proof for the same scaffolding.
If the network can't do it, the forests certainly can't. Moreover, as I see it, your results are decent.
So saying "optimization is a dangerous thing" is like saying "a microscope is a dangerous thing" - it hurts to hit you over the head with it.
Forests, I understand, are used to classify predictors, roughly speaking, not for forecasting :)