Machine learning in trading: theory, models, practice and algo-trading - page 882
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
Binarization has killed a lot of useful information.
Normal woods or random woods, or both?
I put Rattle and R (well, the whole thing glitches...), and now I can not understand how to make comparable settings, as in the screenshot below? Because the standard Rattle settings gave worse results than the program I was using before.
Normal forest or random forest, or both?
In rattle run both forest models called tree and ada. Open the log tab and see the R code, references to the packages used, and you can understand their differences.
I put Rattle and R (what a glitch this whole thing is...),
I don't understand what 's glitchy, lately I've run a huge number of models - everything's fine
And now I can not understand how to make comparable settings, as in the screenshot below? Because sdandartny Rattle settings gave worse results than the progam, which I used before.
The picture from rattle you have unfinished. At a minimum, I should go to the adjacent evaluate tab and see the results there.
But most importantly you need to split the source file into two parts with different names (most likely you will have to do it in R).
On the first file you build all six models and look at their estimates test, validate. Then the name of the second file you enter in the R Dataset field. And on it you get marks again. All the evaluations must be approximately the same!
If these estimates do not coincide, and the results of models on the second file are much worse, then this means that your models are over-trained, and the reason for over-training is the presence of noise (not related to the target variable) predictors.
This is the moment of truth: either you have a set of predictors relevant to a particular target variable or you don't. And no model can fix this sad circumstance. Then begins the stupid work of selecting a pair of "target-predictors", models are not interesting at all, find a pair, then the models are just seeds in R, you will find a dozen in a day and you will make ensembles of them.
PS.
1. Don't listen to forumers who don't know R: their results are not only impossible to repeat, but even understand.
2. There is no problem using R advisor: everything works and is very stable.
3. Do not forget that rattle logs on R all your actions. This protocol can be used as usual R code.
PS.
1. Don't listen to forumers who don't know R: their results are not only impossible to replicate, but even understand.
2. There is no problem using R EA: everything works and is very stable.
3. Do not forget that rattle logs on R all your actions. You can use this protocol as usual code in R
SanSanych, here, I have a question. Do you currently have anything really working and making money? Enough - yes or no.
And if yes, then what does R have to do with it? And if no, then where and what does R have to do with it?
SZY The answer can be guessed with 0.95 probability.)
SanSanych, here, I have a question. Do you currently have anything really working and making money? Is it enough - yes or no.
And if yes, then R is here from what side? And if no, moreover, where and why R is here?
ZZY The answer can be guessed with probability 0.95).
I had financial problems 2 years ago. I solved them in one year. The Expert Advisor that solved these problems had a decision block = random forest + indicators. There were no theoretical guarantees that the future would be like the past.
Financial problems reappeared. Finishing up a new development with a decision block ONLY on R. Don't disclose details except one: there are theoretical justifications and tests to support the theory that the future will be similar to the past. Timing is not clear yet.
Had financial problems two years ago. Solved it in a year. The advisor who solved these problems had a decision unit = random forest + indicators. There were no theoretical guarantees that the future would be like the past.
Financial problems reappeared. Finishing up a new development with a decision block ONLY on R. Don't disclose details except one: there are theoretical justifications and tests to support the theory that the future will be similar to the past. The timeline is not yet clear.
What's interesting here is that you were looking for a way to interact with R as recently as six months to a year ago - the result was the DLL. Before that there was no such possibility at all.
Forests also appeared in your posts about a year ago. Before that it was ADA etc.
The interesting thing here is that you were looking for a way to interact with R as recently as six months or a year ago - as a result, the DLL appeared. Before that there was no such possibility at all.
Forests also appeared in your posts about a year ago. Before that, it was ADA, etc.
Don't you remember the article about scaffolding from 2014? https://www.mql5.com/ru/articles/1165 Or the indicator from 2012? Go to your profile before you make stuff up.
Don't you remember the forest article from 2014? https://www.mql5.com/ru/articles/1165
No, I haven't. I am guided by the MO branch. Here SanSanych was talking about ADA. Then I switched to forests.
But the main thing - the connection between the terminal and the TC is impossible without it. The topic appeared -FormulatingTerms of Reference for MQL4/5 connection with R-2017.01.02 15:59. Ended - 2017.11.30 10:16.
Thus, earlier than 06.2017 no system could be somehow linked to R.
I personally have nothing about SanSanych, he is a very competent and discreet man, doing something of his own unknown, he probably needs R
I prefer python intuitively, although I haven't invented something special to do with it that would be just wow, but I continue to learn it quietly, see if it comes in handy :D
The interesting thing here is that you were looking for a way to interact with R as recently as six months or a year ago - as a result, the DLL appeared. Before that there was no such possibility at all.
I placed R-based indicator in May 2012, i.e. six years ago. Before that I have placed in my kodobase DLR for MT4/R (by 7bit). So the possibility of using R from MT4 EA is a very long time ago.
A year ago I was looking for an implementer who would revise this ancient DLL for 64 bits. Now you can use R from MT5.
Forests also appeared in your posts about a year ago. Before that there were ADA and so on.
I always wrote that I liked ada better, but there was something missing for it in the caret that I used in development and in operation. That's why I used randomForest package in my Expert Advisor.
Fucking miners, sorry Jesus))) You originally did on C4.5. In the rattle you put
a weak tree. Maksimka stoner writes about random wood in general. The wood is for the woods, and the wood is for the woods)))
On your data, file Pred_004_Buy divided in half, head-on can get 0.85.
The data is rubbish and better to throw away. The rest we catch up on our own. In silence...
How do I run this algorithm in Rattle or another R tool, or whatever it is called? Interested in the further use of the results in MT5.
If 0.85 is fucked up, then there should be a 100% result for this algorithm, or is there any weighty argument?