Programmatic

Construct custom configs directly:

from phil import Phil, ImputationConfig, ECTConfig
import pandas as pd

config = ImputationConfig(
  methods=["SimpleImputer", "KNNImputer"],
  modules=["sklearn.impute", "sklearn.impute"],
  grids=[{"strategy": ["mean", "median"]}, {"n_neighbors": [3, 5]}],
)

magic_config = ECTConfig(
  num_thetas=32,
  radius=1.0,
  resolution=64,
  scale=256,
  normalize=True,
  seed=42,
)

model = Phil(samples=20, param_grid=config, config=magic_config)
model.fit(pd.DataFrame({"x": [1, None, 3], "y": [1.0, 2.0, None]}))