experiments.yaml

# general
data_dir: "data"
plot_out_dir: "figures"
path_to_xlemoo: "."

# experiment related
total_iterations: 1000
ml_every_n: [2, 4, 5, 8, 10, 20, 25, 50, 100, 200, 500]
hl_split: [0.05, 0.10, 0.15, 0.20, 0.25, 0.30, 0.35, 0.40, 0.45, 0.50]
runs_per_experiment: 10

# problem configurations start. These must be modified to change the problem the experiments are run with

## carside impact problem
# n_objectives: 4
# n_variables: 7
# problem: "carsideimpact"
# nadir: [40.0, 5.0, 14.0, 30.0]
# ideal: [15.0, 3.0, 10.0, 0.0]
# ref_point: [20.0, 3.5, 11.0, 0.1]
# script_name: "carside_impact.py"

## multiple clutch brakes problem
# n_objectives: 5
# n_variables: 5
# problem: "multipleclutchbrakes"
# nadir: [4.0, 22, 12.0, 120, 1100]
# ideal: [-0.7, 2.9, 1.0, 70.0, 280]
# ref_point: [1.2, 8.2, 3.5, 95.0, 320]
# script_name: "multiple_clutch_brakes.py"

## vehicle crash worthiness
n_objectives: 3
n_variables: 5
problem: "vehiclecrashworthiness"
nadir: [1700.0, 12.0, 0.2]
ideal: [1600.0, 6.0, 0.038]
ref_point: [1650.0, 7.0, 0.05]
script_name: "vehicle_crash_worthiness.py"

# problem configurations end here, the parameters below are related to the evolutionary
# and learning steps of the XLEMOO method.

# LEMOO conf
fitness_fun: "PointMethodASF"
use_darwin: true
use_ml: true
ml_probe: null
ml_threshold: null
darwin_probe: null
darwin_threshold: null

# EA conf
pop_size: 50
population_init_design: "LHSDesign"

# ML conf
h_split: 0.2
l_split: 0.2
instantiation_factor: 10
generation_lookback: 0
ancestral_recall: 0
unique_only: true
ml_iterations_per_cycle: 1

# ML model conf
ml_model_name: "SkopedRuleClassifier"
ml_precision_min: 0.1
ml_n_estimators: 30
ml_max_features: null
ml_max_depth: null
ml_bootstrap: true
ml_bootstrap_features: true

# plotting of heatmaps
plot_for_each_iter: [10, 50, 200, 1000]