params.yaml

seed: 0
sample_count: 3
sub_sample:
  #according_to_columns:
  #  Class_of_Orbit:
  #    - LEO
  #    - GEO
  #  Users:
  #    - Commercial
  N: 1000
parallel:
  # To limit the number of jobs GNU Parallel spawns use add the flag `--jobs n`.
  flags: >-
    --verbose
nullify:
  # Entries added here will be treated as null. For example, to treat "NaN" and
  # "missing" as null uncomment the following two lines:
  # - NaN
  # - missing
schema:
  # Key/value pairs can be added here to explicitly override the inferred schema
  # for each column. The available types are as follows: numerical, nominal, and ignore.
  # For example, to force the "age" column to be treated as a
  # numerical uncomment the following line:
  # age: numerical
  # Apogee_km: numerical
  # Perigee_km: numerical
loom:
  extra_passes: 1
cgpm:
  iterations: 1
  minutes: 1
  #dependence:
  #  # While the CrossCat implementation in CGPM takes dependence constraints, doing so
  #  # throws a not-implemented-error. Hence, we apply a workaround:
  #  #   - Supply a map from a target column name to a list of column names.
  #  #   - Each column in said list gets moved to the target column's view.
  #  Users:
  #    - Purpose
  #  Launch_Site:
  #    - Type_of_Orbit
  ## Independence is un-directional. For two columns foo and bar, setting `foo: - bar`
  ## and `bar: - foo` has the identical effect.
  #independence:
  #  Perigee_km:
  #    - Class_of_Orbit
  #    - Apogee_km
  #  Apogee_km:
  #    - Class_of_Orbit
clojurecat:
  iterations: 150
qc:
  # Set this to the number of synthetic samples desired in QC plots.
  sample_count: 1000
  # Set this to specify the columns visualized in QC plots.
  # When not set, up to 8 columns--taken from the schema--will be visualized.
  # e.g. columns: ["foo", "bar", "baz"]
  columns: null
  # This limits the number of categories shown for categorical variables in
  # 2-dimensional QC plots. Set this to null for no limit.
  category_limit: 10
mi:
  configs:
    # Set MI configs here. If not set, the configs are chosen at random.
    # For example, to set the config for health_status.
    # health_status: ["c) Average", "b) Below average"]
synthetic_data_evaluation:
  # If target is not specified, a random target is chosen for prediction.
  #target: Apogee_km
  predictor: Random_forest # One of "Random_forest" or "GLM"
    #N: 10000 # Subsample held-out dataframe with 1000 samples
  datasets: # Add a list of csv synthetic data files
    - synthetic-data-gensql.csv # this one is generated by default.
    #- synthetic-data-gan.csv
    #- synthetic-data-copula.csv
database:
  table_name: data
  table_path: data/nullified.csv
  model_name: baseline_model
  model_path: data/xcat/xcat.merged.edn
  # model_path: data/sppl/merged.json
  # model_path: data/xcat/complete/sample.0.edn