data_conversion

Data conversion utilities for MMM.

Normalises various input types (pd.DataFrame, xr.Dataset, xr.DataArray) into the canonical xr.Dataset format that the MMM class uses internally.

Public vs. internal variable names

Public name	Internal name	Description
media	_channel	Media channel spend data
target	_target	Response/target variable
_control	_control	Control variable data

The names channel and control cannot be used as data variable names because xarray promotes them to dimension coordinates when the variable and dimension share a name. Use media instead of channel, and use _control for control data.

Examples

import xarray as xr
import pandas as pd
from pymc_marketing.mmm.data_conversion import to_mmm_dataset

# From a DataFrame
df = pd.DataFrame(
    {
        "date": pd.date_range("2025-01-01", periods=10, freq="W"),
        "tv": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
        "digital": [2, 3, 4, 5, 6, 7, 8, 9, 10, 11],
    }
)
y = pd.Series([10, 20, 30, 40, 50, 60, 70, 80, 90, 100], name="sales")

ds = to_mmm_dataset(
    df,
    y,
    date_column="date",
    channel_columns=["tv", "digital"],
    target_column="sales",
)
# ds has variables _channel and _target

Functions

to_mmm_dataset(X[, y, dims, ...])

Normalise X (and optionally y) to a single canonical xr.Dataset.