meridian.data.nd_array_input_data_builder.NDArrayInputDataBuilder

Builds InputData from n-dimensional arrays.

Inherits From: InputDataBuilder

meridian.data.nd_array_input_data_builder.NDArrayInputDataBuilder(
    kpi_type: str
)

controls

frequency

geos

kpi

media

media_spend

media_time_coords

non_media_treatments

organic_frequency

organic_media

organic_reach

population

reach

revenue_per_kpi

rf_spend

time_coords

Methods

build

View source

build() -> meridian.data.input_data.InputData

Builds an InputData.

Constructs an InputData from constituent DataArrays given to this builder thus far after performing one final validation pass over all data arrays for consistency checks.

Returns
A validated InputData.

with_controls

View source

with_controls(
    nd: np.ndarray, control_names: list[str]
) -> 'NDArrayInputDataBuilder'

Reads controls data from a ndarray.

nd must be given with the shape: - (n_geos, n_time, n_controls) - (n_time, n_controls) or (1, n_time, n_controls) for national model.

If called without a call to .geos() first, the data will be assumed to be national-level.

Args
nd The ndarray to read the controls data from.
control_names The names of the control variables.

Returns
The NDArrayInputDataBuilder with the added controls data.

with_kpi

View source

with_kpi(
    nd: np.ndarray
) -> 'NDArrayInputDataBuilder'

Reads KPI data from a ndarray.

nd must be given with the shape: - (n_geos, n_time) - (n_time,) or (1, n_time) for national model.

If called without a call to .geos() first, the data will be assumed to be national-level.

Args
nd The ndarray to read the KPI data from.

Returns
The NDArrayInputDataBuilder with the added KPI data.

with_media

View source

with_media(
    m_nd: np.ndarray, ms_nd: np.ndarray, media_channels: list[str]
) -> 'NDArrayInputDataBuilder'

Reads media and media spend data from the ndarrays.

m_nd must be given with the shape: - (n_geos, n_media_times, n_media_channels) - (n_media_times, n_media_channels) or (1, n_media_times, n_media_channels) for national model.

ms_nd must be given with the shape: - (n_geos, n_times, n_media_channels) - (n_times, n_media_channels) or (1, n_times, n_media_channels) for national model.

If called without a call to .geos() first, the data will be assumed to be national-level.

Args
m_nd The ndarray that contains dimensional media data.
ms_nd The ndarray that contains dimensional media spend data.
media_channels The names of the media channels.

Returns
The NDArrayInputDataBuilder with the added media and media spend data.

with_non_media_treatments

View source

with_non_media_treatments(
    nd: np.ndarray, non_media_channel_names: list[str]
) -> 'NDArrayInputDataBuilder'

Reads non-media treatments data from a ndarray.

nd must be given with the shape: - (n_geos, n_time, n_media_channels) - (n_time, n_media_channels) or (1, n_time, n_media_channels) for national model.

If called without a call to .geos() first, the data will be assumed to be national-level.

Args
nd The ndarray to read the non-media treatments data from.
non_media_channel_names The names of the non-media channels.

Returns
The NDArrayInputDataBuilder with the added non-media treatments data.

with_organic_media

View source

with_organic_media(
    nd: np.ndarray, organic_media_channels: list[str]
) -> 'NDArrayInputDataBuilder'

Reads organic media data from a ndarray.

nd must be given with the shape: - (n_geos, n_media_times, n_organic_media_channels) - (n_media_times, n_organic_media_channels) or (1, n_media_times, n_organic_media_channels) for national model.

If called without a call to .geos() first, the data will be assumed to be national-level.

Args
nd The ndarray to read the organic media data from.
organic_media_channels The names of the organic media channels.

Returns
The NDArrayInputDataBuilder with the added organic media data.

with_organic_reach

View source

with_organic_reach(
    or_nd: np.ndarray, of_nd: np.ndarray, organic_rf_channels: list[str]
) -> 'NDArrayInputDataBuilder'

Reads organic reach and organic frequency data from the ndarrays.

or_nd and of_nd must be given with the shape: - (n_geos, n_media_times, n_organic_rf_channels) - (n_media_times, n_organic_rf_channels) or (1, n_media_times, n_organic_rf_channels) for national model.

If called without a call to .geos() first, the data will be assumed to be national-level.

Args
or_nd The ndarray that contains dimensional reach data.
of_nd The ndarray that contains dimensional frequency data.
organic_rf_channels The names of the organic rf channels.

Returns
The NDArrayInputDataBuilder with the added organic reach and organic frequency data.

with_population

View source

with_population(
    nd: np.ndarray
) -> 'NDArrayInputDataBuilder'

Reads population data from a ndarray.

nd must be given with the shape: - (n_geos,)

If called without a call to .geos() first, the data will be assumed to be national-level.

Args
nd The ndarray to read the population data from.

Returns
The NDArrayInputDataBuilder with the added population data.

with_reach

View source

with_reach(
    r_nd: np.ndarray,
    f_nd: np.ndarray,
    rfs_nd: np.ndarray,
    rf_channels: list[str]
) -> 'NDArrayInputDataBuilder'

Reads reach, frequency, and rf_spend data from the ndarrays.

r_nd and f_nd must be given with the shape: - (n_geos, n_media_times, n_rf_channels) - (n_media_times, n_rf_channels) or (1, n_media_times, n_rf_channels) for national model.

rfs_nd must be given with the shape: - (n_geos, n_times, n_rf_channels) - (n_times, n_rf_channels) or (1, n_times, n_rf_channels) for national model.

If called without a call to .geos() first, the data will be assumed to be national-level.

Args
r_nd The ndarray that contains dimensional reach data.
f_nd The ndarray that contains dimensional frequency data.
rfs_nd The ndarray that contains dimensional rf_spend data.
rf_channels The names of the rf channels.

Returns
The NDArrayInputDataBuilder with the added reach, frequency, and rf_spend data.

with_revenue_per_kpi

View source

with_revenue_per_kpi(
    nd: np.ndarray
) -> 'NDArrayInputDataBuilder'

Reads Revenue per KPI data from a ndarray.

nd must be given with the shape: - (n_geos, n_time) - (n_time,) or (1, n_time) for national model.

If called without a call to .geos() first, the data will be assumed to be national-level.

Args
nd The ndarray to read the Reevenue per KPI data from.

Returns
The NDArrayInputDataBuilder with the added Revenue per KPI data.