Discussing the article: "Mastering ONNX: The Game-Changer for MQL5 Traders" - page 2

[amrhamed83]

@Omega J Msigwa thank you very much for this article. Can you please update the tutorial file since the last version of MALE5 doesn't have CPreprocessing? If not possible, which version of MALE5 was used to run this tutorial?

I am not sure if there is documentation for the library. 

Thanks

[Omega J Msigwa]

amrhamed83 #:

@Omega J Msigwa thank you very much for this article. Can you please update the tutorial file since the last version of MALE5 doesn't have CPreprocessing? If not possible, which version of MALE5 was used to run this tutorial?

I am not sure if there is documentation for the library. 

Thanks

Use the cprecessi mg code attached in this article for anything concerning this tutorial 

[amrhamed83]

Omega J Msigwa #:
Use the cprecessi mg code attached in this article for anything concerning this tutorial 

I am referring to this line : 

 #include <MALE5\preprocessing.mqh> // you point to use https://github.com/MegaJoctan/MALE5/blob/MQL5-ML/preprocessing.mqh which doesn't have CPreprocessing

CPreprocessing<vectorf, matrixf> *norm_x;

[Omega J Msigwa]

Ok got it, if thats the case change that line to:

#include <preprocessing.mqh>

After saving the preprocessing.mqh found on this zip file (attached in the article) , under include folder.

CPreprocessing has been deprecated since v2.0.0. Which is the version used in this article.

Alternatively, call each of the scalers present in the preprocessing file instead of CPreprocessing. Assuming you are using MALE5 version 3.0.0

• RobustScaler
• MinMaxScaler
• StandardizationScaler

Each scaler class provides.

fit_transform( const matrix &X)

Which fits the scaler on the data matrix X and performs the transformation.

transform( const matrix &X)

Which transforms the data matrix X using the fitted scaler.

transform(const vector &X)

Which transforms the data vector X using the fitted scaler.

Let me know if this was helpful.

[big man]

Does the Python model need to be simple for ONNX to work well with MetaTrader 5, or can I use a highly optimized and complex architecture?

[Omega J Msigwa]

big man #:
Does the Python model need to be simple for ONNX to work well with MetaTrader 5, or can I use a highly optimized and complex architecture?

MetaTrader 5 works with models of any complexity.