Machine learning in trading: theory, models, practice and algo-trading - page 1158
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
Is anyone working with anything other than time series/increments?
I need help, I have predictors, I have targets, but I don't know how to get the most out of the data - very little training experience (different models show different results with default settings).
Can someone help on this issue?
What exactly do you want? mt5?
What exactly do you want? mt5?
I want to understand what I can get out of the data, to understand how badly I work with them and what I strive for.
A sample file in csv format.I want to understand what I can get out of the data, so that I can understand how badly I work with it and what I aspire to.
File with the sample in the csv format.So what is it? Classification, regression. Is it necessary to get the error on the trayn and the test?
i.e. how this is supposed to be used in trading
So what is it? Classification, regression. Do you want to get the error on the train and the test?
You can do it both ways :) There is a binary target which can be divided into buying and selling according to the predictor, there is a predictor with 3 targets - buying/selling and no action, and it is possible to represent separately buying/not trading and selling/not trading as a binary and perform either regression or classification.
I would like to get a contingency table, on the test sample and markup model of the test sample, well, that would appear the column with the results of calculating the target.
And a couple of words about how the target was reached.You can do both :) There is a binary target, which can be divided into buy and sell by predictor, there is a predictor with 3 targets - buy/sell and no action, and you can separately buy/no trade and sell/no trade as a binary and do either regression or classification.
I would like to get a contingency table for the test sample and markup with the model of the test sample, so that the column with the results of the target calculation would appear.
And a couple of words about how the goal was achieved.Well if you use mql there will be just 2 errors per test and a trace. Compare them and see to what extent the model generalizes the data on the test one. Error of classification, or logloss, what else is there... RMS if regression
i don't know what a contingency table is
Well, if you use mql there will be just two errors per test and a trace. Compare them and see how the model generalizes the data on the test one. Error of classification, or logloss, what else is there... RMS if regression
I don't know what a contingency table is.
What kind of tools do MQL Alglib forests have?
A conjugation table - like on the picture
And what are the tools of the MQL Alglib forest?
A contingency table is like the table in the picture
Yes, or NS. You can only play with model itself, add trees or neurons.
Yes, or NS. Well, that's essentially what you've already done) the only thing is that you can play with the model itself, add trees or neurons there.
Does it have to work out of the box like this? Isn't there any system for the creation of NS, for example 100 predictors, so you need 10 neurons, i.e. any rules that actually work?
So can you try my dataset?
Is this the way everything should work out of the box? Isn't there any system for building NS, for example 100 predictors, so you need 10 neurons, i.e. any rules that actually work?
So can you try my dataset?
no sure, all by trial and error, usually after 5-10 tries you can pick up the optimum, there are not all so miserable :) for 2 parameter changes
I can, but later ... very much cramped with his new bot :) In general, I threw an example of multiplication table, there your data substitute and all ... but this should deal with filling matrices, as I understand you have not figured out
No, of course, all by trial and error, usually 5-10 tries you can find the optimal one, there are not so bad :) for 2 parameter changes
I can, but later... very much cramped with his new bot :) in general, threw an example of multiplication table, there their data substitute and all... but it should deal with filling matrices, as I understand you have not worked out
Thank you, I'll send it, as you say.