search

PEtab-dev / petab_select

Repository for development of the model selection extension
https://petab-select.readthedocs.io
BSD 3-Clause "New" or "Revised" License
9 stars 0 forks source link

ui method change #24

Closed Doresic closed 2 years ago

Doresic commented 2 years ago

Took most of the functionality of method.py in pypesto into ui.py in petab_select to make petab_select more independent of the calibration tool. The outcome is that the pipeline changed from (all outside petab_select): -> new candidates -> calibrate -> find best ... to now being: -> best of last calibrated candidates (in petab_select) -> new candidates (in petab_select) -> calibrate (outside of petab_select)...

So only the calibration is done outside petab_select. Everything else is done within ui.py.

TODO: Since candidates() in ui.py is changed, all functions/classes that use this function have to be changed accordingly. (to provide the previous_predecessor_model and full history, and to except to receive new candidates which need to be calibrated, history, and local history)

POSSIBLE PROBLEM: pypesto_select_problem.select() in pypesto, i.e. a single selection step is now difficult to adapt, since in a single step the MethodCaller will provide ui.py the initial predecessor_model, ui.py will generate candidates and give them back to the MethodCaller for calibration, but for the best model to be selected the MethodCaller should provide the calibrated candidates to the ui.py again where the best will be selected. So in a single step pypesto_select_problem.select() will only have calibrated candidate models (from the local history). Possible solution: Add the selection of the best model (from the local history provided by the MethodCaller) into pypesto_select_problem.select() after the method_caller() call.

codecov-commenter commented 2 years ago

Codecov Report

:exclamation: No coverage uploaded for pull request base (FAMoS@9922ae3). Click here to learn what that means. The diff coverage is n/a.

@@           Coverage Diff            @@
##             FAMoS      #24   +/-   ##
========================================
  Coverage         ?   77.90%           
========================================
  Files            ?       14           
  Lines            ?     1521           
  Branches         ?        0           
========================================
  Hits             ?     1185           
  Misses           ?      336           
  Partials         ?        0

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

dilpath commented 2 years ago

Add the selection of the best model (from the local history provided by the MethodCaller) into pypesto_select_problem.select() after the method_caller() call.

Resolved as suggested.