Load national-level data

Simulated data is provided as an example for each data type and format in the following sections.

CSV

To load the simulated CSV data using CsvDataLoader:

  1. Map the column names to the variable types. The required variable types are time, controls, kpi, revenue_per_kpi, media and media_spend. For the definition of each variable, see Collect and organize your data.

    coord_to_columns = load.CoordToColumns(
        time='time',
        controls=['GQV', 'Discount', 'Competitor_Sales'],
        kpi='conversions',
        revenue_per_kpi='revenue_per_conversion',
        media=[
            'Channel0_impression',
            'Channel1_impression',
            'Channel2_impression',
            'Channel3_impression',
            'Channel4_impression',
            'Channel5_impression',
        ],
        media_spend=[
            'Channel0_spend',
            'Channel1_spend',
            'Channel2_spend',
            'Channel3_spend',
            'Channel4_spend',
            'Channel5_spend',
        ],
    )
    
  2. Map the media variables and the media spends to the designated channel names that you want to display in the two-page output. In the following example, Channel0_impression and Channel0_spend are connected to the same channel, Channel0.

    correct_media_to_channel = {
        'Channel0_impression': 'Channel0',
        'Channel1_impression': 'Channel1',
        'Channel2_impression': 'Channel2',
        'Channel3_impression': 'Channel3',
        'Channel4_impression': 'Channel4',
        'Channel5_impression': 'Channel5',
    }
    correct_media_spend_to_channel = {
        'Channel0_spend': 'Channel0',
        'Channel1_spend': 'Channel1',
        'Channel2_spend': 'Channel2',
        'Channel3_spend': 'Channel3',
        'Channel4_spend': 'Channel4',
        'Channel5_spend': 'Channel5',
    }
    
  3. Load the data using CsvDataLoader:

    loader = load.CsvDataLoader(
        csv_path=f'/{PATH}/{FILENAME}.csv',
        kpi_type='non_revenue',
        coord_to_columns=coord_to_columns,
        media_to_channel=correct_media_to_channel,
        media_spend_to_channel=correct_media_spend_to_channel,
    )
    data = loader.load()
    

    Where:

    • kpi_type is either 'revenue' or 'non_revenue'.
    • PATH is the path to the data file location.
    • FILENAME is the name of your data file.

Xarray Dataset

To load the simulated Xarray Dataset using XrDatasetDataLoader:

  1. Load the data using pickle:

    import pickle
    with open(f'/{PATH}/{FILENAME}.pkl', 'r') as fh:
      XrDataset=pickle.load(fh)
    

    Where:

    • PATH is the path to the data file location.
    • FILENAME is the name of your data file.
  2. Pass the dataset to XrDatasetDataLoader. Use the name_mapping argument to map the coordinates and arrays. Provide mapping if the names in the input dataset are different from the required names. The required coordinate names are time, control_variable, and media_channel. The required data variables names are kpi, revenue_per_kpi, controls, media, and media_spend.

    loader = load.XrDatasetDataLoader(
        XrDataset,
        kpi_type='non_revenue',
        name_mapping={'channel': 'media_channel',
                      'control': 'control_variable',
                      'conversions': 'kpi',
                      'revenue_per_conversion': 'revenue_per_kpi',
                      'control_value': 'controls',
                      'spend': 'media_spend'},
    )
    
    data = loader.load()
    

    Where:

    • kpi_type is either 'revenue' or 'non_revenue'.

Numpy ndarray

To load numpy ndarrays directly, use NDArrayInputDataBuilder:

  1. Create the data into separate numpy ndarrays.

    import numpy as np
    
    kpi_nd = np.array([[1, 2, 3]])
    controls_nd = np.array([[[1, 2], [3, 4], [5, 6]]])
    revenue_per_kpi_nd = np.array([[1, 2, 3]])
    media_nd = np.array([[[1, 2], [3, 4], [5, 6]]])
    media_spend_nd = np.array([[[1, 2], [3, 4], [5, 6]]])
    
  2. Use a NDArrayInputDataBuilder to set times, as well as give channel or dimension names as required in a Meridian input data. For the definition of each variable, see Collect and organize your data.

    from meridian.data import nd_array_input_data_builder as data_builder
    
    builder = (
        data_builder.NDArrayInputDataBuilder(kpi_type='non_revenue')
    )
    builder.time_coords = ['2024-01-02', '2024-01-03', '2024-01-01']
    builder.media_time_coords = ['2024-01-02', '2024-01-03', '2024-01-01']
    builder = (
      builder
        .with_kpi(kpi_nd)
        .with_revenue_per_kpi(revenue_per_kpi_nd)
        .with_controls(
          controls_nd,
          control_names=["control0", "control1"])
        .with_media(
          m_nd=media_nd,
          ms_nd=media_spend_nd,
          media_channels=["channel0", "channel1"]
        )
    )
    
    data = builder.build()
    

    Where:

    • kpi_type is either 'revenue' or 'non_revenue'.

Pandas DataFrames or other data formats

To load the simulated other data format (such as excel) using DataFrameInputDataBuilder:

  1. Read the data (such as an excel spreadsheet) into one or more Pandas DataFrame(s).

    import pandas as pd
    
    df = pd.read_excel(
        'https://github.com/google/meridian/raw/main/meridian/data/simulated_data/xlsx/national_media.xlsx',
        engine='openpyxl',
    )
    
  2. Use a DataFrameInputDataBuilder to map column names to the variable types required in a Meridian input data. For the definition of each variable, see Collect and organize your data.

    from meridian.data import data_frame_input_data_builder as data_builder
    
    builder = (
        data_builder.DataFrameInputDataBuilder(kpi_type='non_revenue')
            .with_kpi(df, kpi_col="conversions")
            .with_revenue_per_kpi(df, revenue_per_kpi_col="revenue_per_conversion")
            .with_controls(df, control_cols=["GQV", "Discount", "Competitor_Sales"])
    )
    channels = ["Channel0", "Channel1", "Channel2", "Channel3", "Channel4", "Channel5"]
    builder = builder.with_media(
        df,
        media_cols=[f"{channel}_impression" for channel in channels],
        media_spend_cols=[f"{channel}_spend" for channel in channels],
        media_channels=channels,
    )
    
    data = builder.build()
    

    Where:

    • kpi_type is either 'revenue' or 'non_revenue'.

Next, you can create your model.