Do you want to keep or re-detect {{ redetectedSettings }}? {{ recommendKeepSettings ? 'Your previous settings will be lost' : 'The former values might no longer be relevant' }}.

Even if you choose the KEEP SETTINGS option, the following settings will be lost:
  • Treatment variable — cannot be the same as the outcome, it will be set to the previous outcome value '{{ mlTaskDesign.targetVariable }}'
  • Algorithm settings — not the same between classification & regression
  • Architecture — the current code will be commented
  • Metrics — not the same between the different prediction types
  • Weighting strategy — the {{ isCausalPrediction() | targetRoleName}} variable cannot be the sample weight
  • Weighting strategy — regression cannot use class weight
  • Assertions — depend on prediction type & {{ isCausalPrediction() | targetRoleName}}
  • Overrides — depend on prediction type & target
  • Number of skipped time steps (time gap before evaluation) — must be strictly lesser than the forecast horizon