simulate_covariate_data

cross_sectional.simulate.simulate_covariate_data(
    n: int = 1000,
    *,
    n_noise_covariates: int = 0,
    effect_scale: float = 1.0,
    error_var: float = 1.0,
    seed: int | None = None,
)

Simulate the fixed observational setting with covariates.

Parameters

Name	Type	Description	Default
n	`int`	Number of units.	`1000`
n_noise_covariates	`int`	Number of additional noise covariates. The five covariates that enter the design are always generated, so the frame has `5 + n_noise_covariates` covariate columns.	`0`
effect_scale	`float`	Multiplies the treatment effect sizes (the coefficients of `d` in the location and scale components). `1` is the original design and `0` removes the treatment effect.	`1.0`
error_var	`float`	Multiplies the conditional variance of both arms.	`1.0`
seed	`int`	Seed for `numpy.random.default_rng`.	`None`

Returns

Name	Type	Description
	`polars`.`DataFrame`	Columns `x0, ..., x{p-1}`, `treat`, `y` and the potential outcomes `y_0` and `y_1`.