ballet.yml

version: 0.19.5
project:
  project_name: Fragile Families Collaboration
  project_slug: predict-life-outcomes
  package_slug: fragile_families
problem:
  problem_type: regression
  problem_type_details:
    scorer: neg_mean_squared_error
    classification_type:
contrib:
  module_name: fragile_families.features.contrib
  module_path: src/fragile_families/features/contrib
validation:
  project_structure_validator: ballet.validation.project_structure.validator.ProjectStructureValidator
  feature_api_validator: ballet.validation.feature_api.validator.FeatureApiValidator
  feature_accepter:
    name: ballet.validation.feature_acceptance.validator.CompoundAccepter
    params:
      agg: all
      specs:
        - name: ballet.validation.feature_acceptance.validator.VarianceThresholdAccepter
          params:
            threshold: 5e-2
        - name: ballet.validation.feature_acceptance.validator.MutualInformationAccepter
          params:
            threshold: 1e-3
            handle_nan_targets: ignore
  feature_pruner: ballet.validation.feature_pruning.validator.NoOpPruner
  split: leaderboard
github:
  github_owner: ballet
  remote: origin
  pruning_action: no_action
  auto_merge_accepted_features: yes
  auto_close_rejected_features: no
data:
  s3_bucket: mit-dai-ballet-ff
  s3_bucket_region: us-east-1
  splits:
    train: data/train
    leaderboard: data/leaderboard
    test: data/test
  entities_table_name: entities
  targets_table_name: targets
  tables:
    -
      name: entities
      path: entities.parquet
      pandas:
        read_method: read_parquet
        read_kwargs:
          engine: pyarrow
    -
      name: targets
      path: targets.parquet
      pandas:
        read_method: read_parquet
        read_kwargs:
          engine: pyarrow