tea_tasting.config

Global configuration.

get_config(option=None)

Retrieve the current settings of the global configuration.

Parameters:

Name	Type	Description	Default
option	str | None	The option name.	None

Returns:

Type	Description
Any	The specified option value if its name is provided, or a dictionary containing all options otherwise.

Examples:

import tea_tasting as tt


tt.get_config("equal_var")
#> False

Source code in src/tea_tasting/config.py

def get_config(option: str | None = None) -> Any:
    """Retrieve the current settings of the global configuration.

    Args:
        option: The option name.

    Returns:
        The specified option value if its name is provided,
            or a dictionary containing all options otherwise.

    Examples:
        ```python
        import tea_tasting as tt


        tt.get_config("equal_var")
        #> False
        ```
    """
    if option is not None:
        return _global_config[option]
    return _global_config.copy()

set_config(*, alpha=None, alternative=None, confidence_level=None, equal_var=None, n_obs=None, n_resamples=None, power=None, ratio=None, use_t=None, **kwargs)

Update the global configuration with specified settings.

Parameters:

Name	Type	Description	Default
alpha	float | None	Significance level. Default is 0.05.	None
alternative	Literal['two-sided', 'greater', 'less'] | None	Alternative hypothesis. Default is "two-sided".	None
confidence_level	float | None	Confidence level for the confidence interval. Default is 0.95.	None
equal_var	bool | None	Defines whether equal variance is assumed. If True, pooled variance is used for the calculation of the standard error of the difference between two means. Default is False.	None
n_obs	int | Sequence[int] | None	Number of observations in the control and in the treatment together. Default is None.	None
n_resamples	int | None	The number of resamples performed to form the bootstrap distribution of a statistic. Default is 10_000.	None
power	float | None	Statistical power. Default is 0.8.	None
ratio	float | int | None	Ratio of the number of observations in the treatment relative to the control. Default is 1.	None
use_t	bool | None	Defines whether to use the Student's t-distribution (True) or the Normal distribution (False) by default. Default is True.	None
kwargs	Any	User-defined global parameters.	{}

Alternative hypothesis options

• "two-sided": the means are unequal,
• "greater": the mean in the treatment variant is greater than the mean in the control variant,
• "less": the mean in the treatment variant is less than the mean in the control variant.

Examples:

import tea_tasting as tt


tt.set_config(equal_var=True, use_t=False)

experiment = tt.Experiment(
    sessions_per_user=tt.Mean("sessions"),
    orders_per_session=tt.RatioOfMeans("orders", "sessions"),
    orders_per_user=tt.Mean("orders"),
    revenue_per_user=tt.Mean("revenue"),
)

experiment.metrics["orders_per_user"]
#> Mean(value='orders', covariate=None, alternative='two-sided',
#> confidence_level=0.95, equal_var=True, use_t=False)

Source code in src/tea_tasting/config.py

def set_config(
    *,
    alpha: float | None = None,
    alternative: Literal["two-sided", "greater", "less"] | None = None,
    confidence_level: float | None = None,
    equal_var: bool | None = None,
    n_obs: int | Sequence[int] | None = None,
    n_resamples: int | None = None,
    power: float | None = None,
    ratio: float | int | None = None,
    use_t: bool | None = None,
    **kwargs: Any,
) -> None:
    """Update the global configuration with specified settings.

    Args:
        alpha: Significance level. Default is 0.05.
        alternative: Alternative hypothesis. Default is `"two-sided"`.
        confidence_level: Confidence level for the confidence interval.
            Default is `0.95`.
        equal_var: Defines whether equal variance is assumed. If `True`,
            pooled variance is used for the calculation of the standard error
            of the difference between two means. Default is `False`.
        n_obs: Number of observations in the control and in the treatment together.
            Default is `None`.
        n_resamples: The number of resamples performed to form the bootstrap
            distribution of a statistic. Default is `10_000`.
        power: Statistical power. Default is 0.8.
        ratio: Ratio of the number of observations in the treatment
            relative to the control. Default is 1.
        use_t: Defines whether to use the Student's t-distribution (`True`) or
            the Normal distribution (`False`) by default. Default is `True`.
        kwargs: User-defined global parameters.

    Alternative hypothesis options:
        - `"two-sided"`: the means are unequal,
        - `"greater"`: the mean in the treatment variant is greater than the mean
            in the control variant,
        - `"less"`: the mean in the treatment variant is less than the mean
            in the control variant.

    Examples:
        ```python
        import tea_tasting as tt


        tt.set_config(equal_var=True, use_t=False)

        experiment = tt.Experiment(
            sessions_per_user=tt.Mean("sessions"),
            orders_per_session=tt.RatioOfMeans("orders", "sessions"),
            orders_per_user=tt.Mean("orders"),
            revenue_per_user=tt.Mean("revenue"),
        )

        experiment.metrics["orders_per_user"]
        #> Mean(value='orders', covariate=None, alternative='two-sided',
        #> confidence_level=0.95, equal_var=True, use_t=False)
        ```
    """
    params = {k: v for k, v in locals().items() if k != "kwargs"} | kwargs
    for name, value in params.items():
        if value is not None:
            _global_config[name] = tea_tasting.utils.auto_check(value, name)

config_context(*, alpha=None, alternative=None, confidence_level=None, equal_var=None, n_obs=None, n_resamples=None, power=None, ratio=None, use_t=None, **kwargs)

A context manager that temporarily modifies the global configuration.

Parameters:

Name	Type	Description	Default
alpha	float | None	Significance level. Default is 0.05.	None
alternative	Literal['two-sided', 'greater', 'less'] | None	Alternative hypothesis. Default is "two-sided".	None
confidence_level	float | None	Confidence level for the confidence interval. Default is 0.95.	None
equal_var	bool | None	Defines whether equal variance is assumed. If True, pooled variance is used for the calculation of the standard error of the difference between two means. Default is False.	None
n_obs	int | Sequence[int] | None	Number of observations in the control and in the treatment together. Default is None.	None
n_resamples	int | None	The number of resamples performed to form the bootstrap distribution of a statistic. Default is 10_000.	None
power	float | None	Statistical power. Default is 0.8.	None
ratio	float | int | None	Ratio of the number of observations in the treatment relative to the control. Default is 1.	None
use_t	bool | None	Defines whether to use the Student's t-distribution (True) or the Normal distribution (False) by default. Default is True.	None
kwargs	Any	User-defined global parameters.	{}

Alternative hypothesis options

• "two-sided": the means are unequal,
• "greater": the mean in the treatment variant is greater than the mean in the control variant,
• "less": the mean in the treatment variant is less than the mean in the control variant.

Examples:

import tea_tasting as tt


with tt.config_context(equal_var=True, use_t=False):
    experiment = tt.Experiment(
        sessions_per_user=tt.Mean("sessions"),
        orders_per_session=tt.RatioOfMeans("orders", "sessions"),
        orders_per_user=tt.Mean("orders"),
        revenue_per_user=tt.Mean("revenue"),
    )

experiment.metrics["orders_per_user"]
#> Mean(value='orders', covariate=None, alternative='two-sided',
#> confidence_level=0.95, equal_var=True, use_t=False)

Source code in src/tea_tasting/config.py

@contextlib.contextmanager
def config_context(
    *,
    alpha: float | None = None,
    alternative: Literal["two-sided", "greater", "less"] | None = None,
    confidence_level: float | None = None,
    equal_var: bool | None = None,
    n_obs: int | Sequence[int] | None = None,
    n_resamples: int | None = None,
    power: float | None = None,
    ratio: float | int | None = None,
    use_t: bool | None = None,
    **kwargs: Any,
) -> Iterator[Any]:
    """A context manager that temporarily modifies the global configuration.

    Args:
        alpha: Significance level. Default is 0.05.
        alternative: Alternative hypothesis. Default is `"two-sided"`.
        confidence_level: Confidence level for the confidence interval.
            Default is `0.95`.
        equal_var: Defines whether equal variance is assumed. If `True`,
            pooled variance is used for the calculation of the standard error
            of the difference between two means. Default is `False`.
        n_obs: Number of observations in the control and in the treatment together.
            Default is `None`.
        n_resamples: The number of resamples performed to form the bootstrap
            distribution of a statistic. Default is `10_000`.
        power: Statistical power. Default is 0.8.
        ratio: Ratio of the number of observations in the treatment
            relative to the control. Default is 1.
        use_t: Defines whether to use the Student's t-distribution (`True`) or
            the Normal distribution (`False`) by default. Default is `True`.
        kwargs: User-defined global parameters.

    Alternative hypothesis options:
        - `"two-sided"`: the means are unequal,
        - `"greater"`: the mean in the treatment variant is greater than the mean
            in the control variant,
        - `"less"`: the mean in the treatment variant is less than the mean
            in the control variant.

    Examples:
        ```python
        import tea_tasting as tt


        with tt.config_context(equal_var=True, use_t=False):
            experiment = tt.Experiment(
                sessions_per_user=tt.Mean("sessions"),
                orders_per_session=tt.RatioOfMeans("orders", "sessions"),
                orders_per_user=tt.Mean("orders"),
                revenue_per_user=tt.Mean("revenue"),
            )

        experiment.metrics["orders_per_user"]
        #> Mean(value='orders', covariate=None, alternative='two-sided',
        #> confidence_level=0.95, equal_var=True, use_t=False)
        ```
    """
    new_config = {k: v for k, v in locals().items() if k != "kwargs"} | kwargs
    old_config = get_config()
    set_config(**new_config)

    try:
        yield
    finally:
        _global_config.update(**old_config)