Proposed feature There are datasets where there is

Dear <a class="user-mention notranslate" data-hovercard-type="user" data-hovercard-url

Thank you <a class="user-mention notranslate" data-hovercard-type="user" data-hovercar

Lonely PSUs about samplics HOT 2 CLOSED

krosenfeld-IDM commented on May 23, 2024

Lonely PSUs

from samplics.

Comments (2)

MamadouSDiallo commented on May 23, 2024 1

Dear @krosenfeld-IDM

You can try single_psu and/or strata_comb to see if that resolve your issue.

def estimate(
        self,
        y: Array,
        samp_weight: Optional[Array] = None,
        x: Optional[Array] = None,
        stratum: Optional[Series] = None,
        psu: Optional[Series] = None,
        ssu: Optional[Series] = None,
        domain: Optional[Series] = None,
        by: Optional[Series] = None,
        fpc: Union[dict[StringNumber, Number], Series, Number] = 1.0,
        deff: bool = False,
        coef_var: bool = False,
        single_psu: Union[SinglePSUEst, dict[StringNumber, SinglePSUEst]] = SinglePSUEst.error,
        strata_comb: Optional[dict[Array, Array]] = None,
        as_factor: bool = False,
        remove_nan: bool = False,
    ) -> None:

To use single_psu, you will import SinglePSUEst. It is an enum and you can decide to skip the stratum, treat it as a certainty stratum, or combine strata to avoid single psu issue. If you decide to combine, you can use strata_combine to indicate the stata you want to combine.

Check these tests for examples on how to use the API: https://github.com/samplics-org/samplics/blob/main/tests/estimation/test_single_psu_strata.py

Note that I have not documented these use cases because I am re-writting the package and the APIs will change significantly.

With regards

from samplics.

krosenfeld-IDM commented on May 23, 2024

Thank you @MamadouSDiallo - this is a good solution. I'll keep an eye out for the next release.

from samplics.

Recommend Projects