pflacco.sampling#

pflacco.sampling.create_initial_sample(dim: int, n: Optional[int] = None, sample_coefficient: int = 50, lower_bound: Union[List[float], float] = 0, upper_bound: Union[List[float], float] = 1, sample_type: str = 'lhs', seed: Optional[int] = None) → DataFrame#

Sampling of the decision space.

Parameters:

  • dim (int) – Dimensionality of the search space.

  • n (Optional[int], optional) – Fixed number of samples to create. In ELA, this is typically scaled to the dimensionalty of the problem, e.g., n=50*dim, by default None.

  • sample_coefficient (int, optional) – Factor which is used to determine the sample size in conjuction with the problem dimensionality, by default 50.

  • lower_bound (Union[List[float], float], optional) – Lower bound of variables of the decision space, by default 0.

  • upper_bound (Union[List[float], float], optional) – Upper bound of variables of the decision space, by default 1.

  • sample_type (str, optional) – Type of sampling strategy. Should be one of (‘lhs’, ‘random’, ‘sobol’), by default ‘lhs’.

  • seed (Optional[int], optional) – Seed for reproducability, by default None

Returns:

n x dim shaped Pandas dataframe containing the different samples.

Return type:

pd.DataFrame