Discussion of article "CatBoost machine learning algorithm from Yandex with no Python or R knowledge required" - page 7
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
Hello.
If we are talking about replacing an indicator, it is possible to train it, but if we are talking about predicting extrema, it is much more difficult, but the CatBoost algorithm allows you to do it, if there are corresponding predictors.
So far no one has solved this problem...
Or they have, but they are prudently silent about the solution.
Thanks very much for sharing this great article and sample codes, while I have problem Running "_01_Train_All.bat" ,
scrrenshot as below:
Files in the folder:
could you please help to figure out where is the problem, or any step misconfigured?
very appreciated for your kindly support!
Thanks very much for sharing this great article and sample codes, while I have problem Running "_01_Train_All.bat" ,
scrrenshot as below:
Files in the folder:
could you please help to figure out where is the problem, or any step misconfigured?
very appreciated for your kindly support!
For those who did not understand some of the described points on setup and launch, set out in the article, I decided to make a video on the example of EURUSD - watch with subtitles(!).
Huge amount of work and time spent. It's time to start using specialised languages (R/Python/Julia, etc.). They give incomparably more freedom of choice and considerable time saving in development.
Otherwise, good luck in your difficult choice.
Huge amount of work and time spent. It's time to start using specialised languages (R/Python/Julia, etc.). They give incomparably more freedom of choice and considerable time saving in development.
Otherwise, good luck in your difficult choice.
Indeed, the amount of work has been done a lot, and when it is now all automated, the time to learn everything takes not much and you don't need to learn an additional language (R/Python/Julia, etc.) and spend years of your life on it, but already now use the available powerful tool for MO. In any case, knowing a language is too little to achieve results - you need to be able to use its potential, libraries and have the appropriate education. In my case it is more convenient for me to come up with my own idea and do everything the way I see it, and maybe it has already been invented before me, but I just don't know it. Besides, the learning process itself will be faster without the JA and it is easy to transfer it to other computers for parallel data enumeration without organising a complex infrastructure - just divide the bat file into the necessary number of parts according to the number of machines, and do not install packages and settings for learning everywhere - it saves a lot of time.
As a result - for experiments, with deep knowledge of the set of analytical tools and appropriate education will be optimally suited to NPS, and for pipeline, fast and stable, my method will be better.
Alexei thank you very much for the article. Great!
Where could I read Catboost manual in Russian? Google and Yandex offer only in English, and it is not very good with it (translator does not translate terms very well).
If to learn the settings, it would be possible to twist parameters....
And also how to build diagrams as you have on the screens?
If to learn the settings, it would be possible to twist the parameters....
And also how to build diagrams as you have on the screens?
Glad you liked the article!
Unfortunately Yandex puts out all information only in English, the meaning of most parameters is told in the video that is in the article, besides I selected the most effective influencing on training and put them in the script " CB_Bat" - gave a description in the article. It is better to be concerned about the creation of new predictors - it can qualitatively improve the model, twisting the settings of the model - it is often a shift of the random in one direction or another. Experiment and you will understand which parameter is important and which is not so important.
The charts were built in Excel.
Here is a playlist on CatBoost/Hello, there were errors when compiling
I am downloading files (zip) and compiled - all work.