mlr3tuning (version 0.1.0)

TerminatorStagnation: Terminator that stops when tuning does not improve

Description

Class to terminate the tuning after the performance stagnates, i.e. does not improve more than threshold over the last iters iterations.

Arguments

Format

R6::R6Class object inheriting from Terminator.

Construction

t = TerminatorStagnation$new()

Parameters

  • iters :: integer(1) Number of iterations to evaluate the performance improvement on, default is 10.

  • threshold :: numeric(1) If the improvement is less than threshold, tuning is stopped, default is 0.

See Also

Other Terminator: TerminatorClockTime, TerminatorCombo, TerminatorEvals, TerminatorModelTime, TerminatorNone, TerminatorPerfReached, Terminator, mlr_terminators

Examples

Run this code
# NOT RUN {
TerminatorStagnation$new()
term("stagnation", iters = 5, threshold = 1e-5)
# }

Run the code above in your browser using DataLab