Machine learning in trading: theory, models, practice and algo-trading - page 933
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
This is in the R script, which I showed Michael about a hundred pages ago. The genetic algorithm tries parameters for elmnn (activation function, gpsh grain, number of hidden neurons). The fitness function for genetics trains a committee of elmnn models using these parameters, evaluated through kfold, etc.
I wrote this script myself, when I was inspired by your article about elmnn and Bayesian optimization. But made genetics instead of baes, it works much faster for me, and I made committee evaluation to my taste.
Understood. You didn't write about this before or I missed it...
Good luck
The fitness function for genetics trains a committee of elmnn models using these parameters, evaluated through kfold, etc.
Can you tell me more about training through the fitness function? Can my predictors work for this type of NS? How to try if they can?
Over what period of time do you dump the data?
I have a gap of 2 years and the difference between the data is 15 seconds. Predictors: 30 natural and over 1000 generated in "(double)(val1 < val2)" format.
At first I also considered that the number of predictors must be reduced, but practice showed that more is better.
Of course, 1000 predictors in 2 years gives me about 3GB. Using R for such volumes is not serious.
I wonder why?
Python beat R in datamining because there is Cython and Jython hooked up to projects like TensorFlow, Spark, MXNet...
Python hasn't beaten anyone. Did you see who joined the R consortium? Don't start another pointless argument which language is better. You can use both at the same time. The part about Cython and Jython is hilarious.
Good luck
Anyone can be part of the consortium, but that does not mean full support.
RStudio, unlike Anaconda, does not know how to handle graphs properly: scale, coordinates, etc.
It is https://jupyter.org/ that is now the standard for server-side computation.
Can you elaborate on the training through the fitness function? Can my predictors be useful for this type of NS? How to try, if they can?
Roughly speaking, there is a function that has parameters - neuronka settings. The function trains the neuronka according to these parameters, and evaluates the result. And the genetic optimizer selects all of these function parameters to get a higher score, i.e. it calls this function with different parameters, thousands of times, trying to find a better setting. This is similar to the genetics in the mt5 optimizer.
You can try it, I have an example R code in my blog, but you have to add evaluation and sifting of predictors there, the neuron itself won't do it. And without pruning of predictors most likely you'll get overfits and bad results.
Roughly speaking, there is a function that has parameters - neuronka settings. The function trains the neural network according to these parameters, and evaluates the result. The genetic optimizer selects all these function parameters to get a higher score, i.e. it calls this function with different parameters, thousands of times, trying to find a better setting. This is similar to the genetics in the mt5 optimizer.
You can try it, I have an example R code in my blog, but you have to add evaluation and sifting of predictors there, the neuron itself won't do it. And without sifting of predictors most likely there will be overfit and bad result.
Yeah, I'll put it in my queue - I wanted to test the scaffolding first.
But my question is, why predictors can't be sifted in optimizer?
Anyone can be part of the consortium, but that does not mean full support.
RStudio, unlike Anaconda, does not know how to handle graphs properly: scale, coordinates, etc.
It is https://jupyter.org/ that is now the standard for server-side computing.
Funding development, implementation in their products is not full support?
You're building your sentence wrong. Write like this: "I don't know how to work with charts properly in Rstudio, I haven't learned".
The rest without comment, very naive.
And let's stop this argument. Write specifically on the MO. You have experience, though there are no positive results, but that will come with time. Tell us about them, share your experiences. It will be much more useful than just ranting about one language being better than another.
Good luck
You're not making the right sentence. Write like this: "I don't know how to work properly with graphs in Rstudio, I haven't learned.
And there are no graphs in R itself. God knows what through graphing packages too.
Filter_02 2016 arr_Buy
There class "1" even exceeds "0" in number, so there are less false inputs compared to before. Try this tree in your EA please? I wonder what will be on the profit chart myself.
30715182167207681753However, the question is, why can't the predictors be enumerated in the optimizer?
You can