eds_scikit.biology.viz.aggregate

aggregate_measurement

aggregate_measurement(measurement: DataFrame, stats_only: bool, overall_only: bool, value_column: str, unit_column: str, category_columns = [], debug = False)

Aggregates measurement dataframe in three descriptive and synthetic dataframe : - measurement_stats - measurement_volumetry - measurement_distribution

Useful function before plotting.

PARAMETER	DESCRIPTION
measurement	description TYPE: DataFrame
stats_only	description TYPE: bool
overall_only	description TYPE: bool
category_columns	description, by default [] TYPE: list, optional DEFAULT: []

RETURNS	DESCRIPTION
_type_	description

Source code in eds_scikit/biology/viz/aggregate.py

def aggregate_measurement(
    measurement: DataFrame,
    stats_only: bool,
    overall_only: bool,
    value_column: str,
    unit_column: str,
    category_columns=[],
    debug=False,
):
    """Aggregates measurement dataframe in three descriptive and synthetic dataframe :
      - measurement_stats
      - measurement_volumetry
      - measurement_distribution

    Useful function before plotting.

    Parameters
    ----------
    measurement : DataFrame
        _description_
    stats_only : bool
        _description_
    overall_only : bool
        _description_
    category_columns : list, optional
        _description_, by default []

    Returns
    -------
    _type_
        _description_
    """

    check_columns(
        df=measurement,
        required_columns=[
            "measurement_id",
            unit_column,
            "measurement_date",
            value_column,
        ]
        + category_columns,
        df_name="measurement",
    )

    measurement.shape

    # Truncate date
    measurement["measurement_month"] = (
        measurement["measurement_date"].astype("datetime64").dt.strftime("%Y-%m")
    )
    measurement = measurement.drop(columns=["measurement_date"])

    # Filter measurement with missing values
    filtered_measurement, missing_value = filter_missing_values(measurement)

    # Compute measurement statistics by code
    measurement_stats = _describe_measurement_by_code(
        filtered_measurement,
        overall_only,
        value_column,
        unit_column,
        category_columns,
        debug,
    )

    if stats_only:
        return {"measurement_stats": measurement_stats}

    # Count measurement by care_site and by code per each month
    measurement_volumetry = _count_measurement_by_category_and_code_per_month(
        filtered_measurement,
        missing_value,
        value_column,
        unit_column,
        category_columns,
        debug,
    )

    # Bin measurement values by care_site and by code
    measurement_distribution = _bin_measurement_value_by_category_and_code(
        filtered_measurement, value_column, unit_column, category_columns, debug
    )

    return {
        "measurement_stats": measurement_stats,
        "measurement_volumetry": measurement_volumetry,
        "measurement_distribution": measurement_distribution,
    }

add_mad_minmax

add_mad_minmax(measurement: DataFrame, category_cols: List[str], value_column: str = 'value_as_number', unit_column: str = 'unit_source_value') -> DataFrame

Add min_value, max_value column to measurement based on MAD criteria.

PARAMETER	DESCRIPTION
measurement	measurement dataframe TYPE: DataFrame
category_cols	measurement category columns to perform the groupby on when computing MAD TYPE: List[str]
value_column	measurement value column on which MAD will be computed TYPE: str DEFAULT: 'value_as_number'

RETURNS	DESCRIPTION
DataFrame	measurement dataframe with added columns min_value, max_value

Source code in eds_scikit/biology/viz/aggregate.py

def add_mad_minmax(
    measurement: DataFrame,
    category_cols: List[str],
    value_column: str = "value_as_number",
    unit_column: str = "unit_source_value",
) -> DataFrame:
    """Add min_value, max_value column to measurement based on MAD criteria.

    Parameters
    ----------
    measurement : DataFrame
        measurement dataframe
    category_cols : List[str]
        measurement category columns to perform the groupby on when computing MAD
    value_column : str
        measurement value column on which MAD will be computed

    Returns
    -------
    DataFrame
        measurement dataframe with added columns min_value, max_value
    """
    measurement_median = (
        measurement[category_cols + [value_column]]
        .groupby(
            category_cols,
            as_index=False,
            dropna=False,
        )
        .median()
        .rename(columns={value_column: "median"})
    )

    # Add median column to the measurement table
    measurement_median = measurement_median.merge(
        measurement[
            category_cols
            + [
                value_column,
            ]
        ],
        on=category_cols,
    )

    # Compute median deviation for each measurement
    measurement_median["median_deviation"] = abs(
        measurement_median["median"] - measurement_median[value_column]
    )

    # Compute MAD per care site and code
    measurement_mad = (
        measurement_median[
            category_cols
            + [
                "median",
                "median_deviation",
            ]
        ]
        .groupby(
            category_cols
            + [
                "median",
            ],
            as_index=False,
            dropna=False,
        )
        .median()
        .rename(columns={"median_deviation": "MAD"})
    )

    measurement_mad["MAD"] = 1.48 * measurement_mad["MAD"]

    # Add MAD column to the measurement table
    measurement = measurement_mad.merge(
        measurement,
        on=category_cols,
    )

    # Compute binned value
    measurement["max_value"] = measurement["median"] + 4 * measurement["MAD"]
    measurement["min_value"] = measurement["median"] - 4 * measurement["MAD"]

    return measurement