RemixAutoML (version 0.5.0)

RL_ML_Update: RL_ML_Update

Description

RL_ML_Update updates the bandit probabilities for selecting different grids

Usage

RL_ML_Update(
  ModelType = "classification",
  grid_eval_metric = grid_eval_metric.,
  Iteration = counter,
  NewGrid. = NewGrid,
  NewPerformance. = NewPerformance,
  BestPerformance. = BestPerformance,
  Trials. = Trials,
  Successes. = Successes,
  GridIDs. = GridIDs,
  BanditArmsN. = BanditArmsN,
  RunsWithoutNewWinner. = RunsWithoutNewWinner,
  MaxRunsWithoutNewWinner. = MaxRunsWithoutNewWinner,
  MaxModelsInGrid. = MaxModelsInGrid,
  MaxRunMinutes. = MaxRunMinutes,
  TotalRunTime. = TotalRunTime,
  BanditProbs. = BanditProbs
)

Arguments

ModelType

"classification", "regression", and "multiclass"

grid_eval_metric

grid_eval_metric.

Iteration

Model iteration number

NewGrid.

Previous grid passed in

NewPerformance.

Internal

BestPerformance.

Internal

Trials.

Numeric vector with the total trials for each arm

Successes.

Numeric vector with the total successes for each arm

GridIDs.

The numeric vector that identifies which grid is which

BanditArmsN.

The number of arms in the bandit

RunsWithoutNewWinner.

Counter of the number of models previously built without being a new winner

MaxRunsWithoutNewWinner.

Maximum number of models built without a new best model (constraint)

MaxModelsInGrid.

Maximum number of models to build (constraint)

MaxRunMinutes.

Run time constraint

TotalRunTime.

Cumulative run time in minutes

BanditProbs.

Inital probabilities from RL_Initialize()

See Also

Other Reinforcement Learning: CatBoostGridTuner(), RL_Initialize(), RL_Update(), RPM_Binomial_Bandit(), XGBoostGridTuner()

Examples

# NOT RUN {
RL_Update_Output <- RL_ML_Update(
  ModelType = "classification",
  grid_eval_metric = grid_eval_metric.,
  Iteration = run,
  NewGrid. = NewGrid,
  NewPerformance. = NewPerformance,
  BestPerformance. = BestPerformance,
  Trials. = Trials,
  Successes. = Successes,
  GridIDs. = GridIDs,
  BanditArmsN. = BanditArmsN,
  RunsWithoutNewWinner. = RunsWithoutNewWinner,
  MaxRunsWithoutNewWinner. = MaxRunsWithoutNewWinner,
  MaxNumberModels. = MaxNumberModels,
  MaxRunMinutes. = MaxRunMinutes,
  TotalRunTime. = TotalRunTime,
  BanditProbs. = BanditProbs)
BanditProbs <- RL_Update_Output[["BanditProbs"]]
Trials <- RL_Update_Output[["Trials"]]
Successes <- RL_Update_Output[["Successes"]]
NewGrid <- RL_Update_Output[["NewGrid"]]
# }