Skip to content

No hyperparameter optimization example/tutorial incorporating warmstart  #34

Open
@cshih14

Description

@cshih14

Description

Currently, there is no example or demonstration of a hyperparameter optimization function that uses warmstart (GridSearchCV and RandomSearchCV ignore warmstart) and tunes multiple classifier types in the same function. For comparisons between algorithm performance on multiple datasets, a common experiment for tasks such as benchmarking, such a function would be especially useful and reduce computational time.

Enhancement

  • I will write a function that, given a classifier and a dictionary of two hyperparameters, finds the optimal pair of hyperparameter values using grid search. In this specific example, max_features and n_estimators will be the parameters being tuned.

  • The function will implement warmstart for searching through increasing numbers of estimators in the ensemble.

  • A function will be included that visualizes the performance of each parameter value pair in a heatmap to show more detail on how the optimal parameters were determined.

Metadata

Metadata

Assignees

Labels

documentationImprovements or additions to documentation

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions