Skip to content

Adaptive Amazon-kind Simulation - blend Predictiveness and Anti-biased #267

Description

@akolonin

Once #266 and #247 are done, need to explore:

  1. Why Predictiveness works not good enough compared to "non-adaptive simulation"?
  2. Why blending Anti-biased and Predictiveness provides results worse than each of them alone?
  3. If it is because of different impacts of "conservatism" metric (which seems to be the case with "non-adaptive" simulation), conflicting when blending, add "predictiveness_conservatism" (maybe invent shorter name of the parameter) for "predictiveness" alone.

Metadata

Metadata

Assignees

Labels

Type

No type
No fields configured for issues without a type.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions