Model Data Packages
Polaris is a first-of-its-kind incrementality measurement product. The insights we provide through a powerful blend of media mix modeling (MMM) and experimentation are incredibly valuable. That said, those insights, in the form of well-understood KPIs, are only as good as the model. Therefore, we feel it is important to supply model information so the insights can be better understood. It is by no means mandatory to understand the model information in its entirety or at all, and it does require some statistical knowledge, but it is always available.
It is important to understand that the MMM is a hierarchical model. We provide information about each level in the model where applicable, which is why you may see a “level” column as well as why you may see a row for the “Facebook” channel and another row for the “Facebook” channel and “Campaign A” campaign. The model may fit Facebook in a channel level that acts as the parent level to the Facebook campaigns.
Furthermore, it should be noted that separate models are trained for each metric. That is why you’ll notice a metric column in many model data files. For example, a campaign can have a different coefficient for installs than it does for revenue day 1.
Most model data files contain the following basic columns. Some contain additional columns and some only contain a subset of these basic columns.
There are 4 different model data files in each data package. Click each for more information.
Components
Coefficients
Collinearities
Errors
Basics
It is important to understand that the MMM is a hierarchical model. We provide information about each level in the model where applicable, which is why you may see a “level” column as well as why you may see a row for the “Facebook” channel and another row for the “Facebook” channel and “Campaign A” campaign. The model may fit Facebook in a channel level that acts as the parent level to the Facebook campaigns.
Furthermore, it should be noted that separate models are trained for each metric. That is why you’ll notice a metric column in many model data files. For example, a campaign can have a different coefficient for installs than it does for revenue day 1.
Columns
Most model data files contain the following basic columns. Some contain additional columns and some only contain a subset of these basic columns.
| Name | Description | Example |
|---|---|---|
| date | The UTC date of the row in ISO format. | 2021-01-31 |
| channel | The channel of the row. If blank, the row represents all channels. | |
| campaign | The campaign of the row. If blank, the row represents all campaigns. | Campaign_A |
| country | The country of the row. It is never blank unless a blank country was referenced in the input. | US |
| site | The site of the row. If blank, the row represents all sites. | Site_123 |
| level | The model level of the row. This column references the internal model hierarchy. | site |
| metric | The metric of the row. | installs |
Files
There are 4 different model data files in each data package. Click each for more information.
Components
Coefficients
Collinearities
Errors
Updated on: 28/07/2022
Thank you!