ONNX usage on MQL5
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
Currently I'm coding AI stragey program on MQL5.
1) Train a data on python with lightGBM binary
This training data works well on python.
Prediction results are appropriate numbers.
2) Convert it to Onnx format with onnxmltools
However, the training data of onnx on MQL5 returns messed up numbers.
There are 2 nodes.
Case1:
The node[label] should be 0 or 1. This is binary.
But, it returns like "1.4636834311774793e-38".
This is neither 0 nor 1.
[label]
1.4636834311774793e-38
[probabilites]
0.0
Case2:
The node[probabilities] should be from 0 to 1 numbers.
This resulty is also strange.
When operating in python, numbers close to 0 do not appear.
The numbers are clustered between 0.4 and 0.6.
[label]
0.0
[probabilites]
2.386649525030901e+26
--------------
Since I am training in binary, even if there is a bug in the program or a mistake in the data given, the [label] should be 0 or 1.
Could someone please advise me on this matter?