balance.stats_and_plots.impact_of_weights_on_outcome

balance.stats_and_plots.impact_of_weights_on_outcome.compare_adjusted_weighted_outcome_ss(adjusted0: Sample, adjusted1: Sample, method: str = 't_test', conf_level: float = 0.95, round_ndigits: int | None = 3) → pd.DataFrame

Compare two adjusted Samples by testing outcomes under each set of weights.

Parameters:

• adjusted0 – First adjusted Sample (w0).

• adjusted1 – Second adjusted Sample (w1).

• method – Statistical test to use (“t_test”).

• conf_level – Confidence level for the mean difference interval.

• round_ndigits – Optional rounding for numeric outputs.

Returns:

Outcome-by-statistic table comparing weighted outcomes.

Return type:

pd.DataFrame

Examples: .. code-block:: python

import pandas as pd

from balance.sample_class import Sample from balance.stats_and_plots.impact_of_weights_on_outcome import (

compare_adjusted_weighted_outcome_ss,

)

sample = Sample.from_frame(

pd.DataFrame(

{

“id”: [1, 2, 3], “x”: [0.1, 0.2, 0.3], “weight”: [1.0, 1.0, 1.0], “outcome”: [1.0, 2.0, 3.0],

}

), id_column=”id”, weight_column=”weight”, outcome_columns=(“outcome”,),

) target = Sample.from_frame(

pd.DataFrame(

{

“id”: [4, 5, 6], “x”: [0.1, 0.2, 0.3], “weight”: [1.0, 1.0, 1.0], “outcome”: [1.0, 2.0, 3.0],

}

), id_column=”id”, weight_column=”weight”, outcome_columns=(“outcome”,),

) adjusted_a = sample.set_target(target).adjust(method=”null”) adjusted_b = sample.set_target(target).adjust(method=”null”) adjusted_b.set_weights(pd.Series([1.0, 2.0, 3.0], index=adjusted_b.df.index))

impact = compare_adjusted_weighted_outcome_ss(

adjusted_a, adjusted_b, round_ndigits=3

) print(impact.to_string())

mean_yw0  mean_yw1  mean_diff  diff_ci_lower  diff_ci_upper  t_stat  p_value    n

outcome outcome 2.0 4.667 2.667 -4.922 10.256 1.512 0.27 3.0

balance.stats_and_plots.impact_of_weights_on_outcome.weights_impact_on_outcome_ss(y: Iterable[float] | pd.Series | np.ndarray, w0: Iterable[float] | pd.Series | np.ndarray, w1: Iterable[float] | pd.Series | np.ndarray, method: str = 't_test', conf_level: float = 0.95) → pd.Series

Evaluate whether weighting changes the outcome by testing y*w0 vs y*w1.

Note: Weights are normalized to have mean 1 before computing the weighted products. In the balance package, weights are typically normalized to sum to the sample size, so this additional normalization ensures comparability.

Parameters:

• y – Outcome values.

• w0 – Baseline weights.

• w1 – Alternative weights.

• method – Statistical test to use (“t_test”).

• conf_level – Confidence level for the mean difference interval.

Returns:

Summary statistics for the weighted outcome comparison.

Return type:

pd.Series

Examples: .. code-block:: python

import pandas as pd

from balance.stats_and_plots.impact_of_weights_on_outcome import (

weights_impact_on_outcome_ss,

)

result = weights_impact_on_outcome_ss(

y=pd.Series([1.0, 2.0, 3.0, 4.0]), w0=pd.Series([1.0, 1.0, 1.0, 1.0]), w1=pd.Series([1.0, 2.0, 1.0, 2.0]), method=”t_test”,

) print(result.round(3).to_string())

mean_yw0         2.500
mean_yw1         4.000
mean_diff        1.500
diff_ci_lower   -1.547
diff_ci_upper    4.547
t_stat           1.567
p_value          0.215
n                4.000