Source code for synthyverse.generators.synthpop_generator.synth_pop

import pandas as pd
import numpy as np
from synthpop import DataProcessor, MissingDataHandler

from .synthpop_dir.cart import CARTMethod

from ..base import TabularBaseGenerator




class SynthpopGenerator(TabularBaseGenerator):
    """From the popular Synthpop R package.

    Synthpop uses CART (Classification and Regression Trees) to model conditional marginal distributions.
    Synthetic data is autoregressively generated by sampling from leaf nodes.

    Uses the python-synthpop pypi package implementation.

    Paper: "synthpop: Bespoke creation of synthetic data in R" by Nowok et al. (2016).

    Args:
        smoothing (bool): Whether to use smoothing for continuous variables. Default: False.
        proper (bool): Whether to apply a resampling (proper) step during fitting. Default: False.
        minibucket (int): Minimum samples in the leaf nodes. Increase to reduce overfitting. Default: 5.
        tree_params (dict): Dictionary of additional parameters for tree construction (scikit-learn decision trees). Default: {}.
        random_state (int): Random seed for reproducibility. Default: 0.
        **kwargs: Additional arguments passed to TabularBaseGenerator.

    Example:
        >>> import pandas as pd
        >>> from synthyverse.generators import SynthpopGenerator
        >>>
        >>> # Load data
        >>> X = pd.read_csv("data.csv")
        >>> discrete_features = ["category_col"]
        >>>
        >>> # Create generator
        >>> generator = SynthpopGenerator(
        ...     smoothing=True,
        ...     proper=True,
        ...     minibucket=5,
        ...     random_state=42
        ... )
        >>>
        >>> # Fit and generate
        >>> generator.fit(X, discrete_features)
        >>> X_syn = generator.generate(1000)
    """

    name = "synthpop"

    def __init__(
        self,
        smoothing: bool = False,
        proper: bool = False,
        minibucket: int = 5,
        tree_params: dict = {},
        random_state: int = 0,
        **kwargs,
    ):
        super().__init__(random_state=random_state, **kwargs)
        self.smoothing = smoothing
        self.proper = proper
        self.minibucket = minibucket
        self.random_state = random_state
        self.tree_params = tree_params

    def _fit_model(
        self, X: pd.DataFrame, discrete_features: list, X_val: pd.DataFrame = None
    ):
        metadata = MissingDataHandler().get_column_dtypes(X)
        self.processor = DataProcessor(metadata)
        x_pr = self.processor.preprocess(X)

        self.model = CARTMethod(
            metadata,
            smoothing=self.smoothing,
            proper=self.proper,
            minibucket=self.minibucket,
            tree_params=self.tree_params,
            random_state=self.random_state,
        )
        self.model.fit(x_pr)

    def _generate_data(self, n: int):
        syn = self.model.sample(n)
        syn = self.processor.postprocess(syn)
        return syn