search

oceanprotocol / pdr-backend

Instructions & code to run predictoors, traders, more.
Apache License 2.0
22 stars 15 forks source link

[Sim] Low profit if regressor models used #1293

Open graceful-coder opened 4 days ago

graceful-coder commented 4 days ago

Describe the bug Multisim results show very low values for pdr_profit_OCEAN and trader_profit_USD when testing RegrLinearLS & RegrLinearRidge models with the classifiers: None, Isotonic & Sigmoid.

To Reproduce Steps to reproduce the behavior:

  1. Do a fresh pdr-backend clone/install including the lastest Jun 22nd commit that also includes the fixes for https://github.com/oceanprotocol/pdr-backend/issues/1279, https://github.com/oceanprotocol/pdr-backend/issues/1267, https://github.com/oceanprotocol/pdr-backend/pull/1212, https://github.com/oceanprotocol/pdr-backend/issues/1269, & https://github.com/oceanprotocol/pdr-backend/pull/1275.
  2. Modify my_ppss.yaml to test both RegrLinearLS & RegrLinearRidge model approaches with each classifier (None, Isotonic, Sigmoid). You will need 6 different versions of my_ppss.yaml, one for each model & classifier.
  3. Download and view the CSV results for RegrLinearLS & RegrLinearRidge to verify that the pdr_profit_OCEAN & trader_profit_USD values are extremely low values.
  4. Compare that the RegrLinearLS results are similar to the RegrLinearLS data on Gdrive here.
  5. Compare that the RegrLinearRidge results are similar to the RegrLinearRidge data on Gdrive here.

Expected behavior < +/-5000 pdr_profit_OCEAN & < +/-500 trader_profit_USD values for multisim tests of ClassifLinearRidge using all 3 classifiers.