How to manage weather data in the Meridian?

[ramadhita-ind]

Hi! I’m new to this modeling. I have weather data at a 4-hour granularity, meaning there are four data points per day. How can I incorporate this as a control variable in the model, given that the recommended data level is weekly? I feel that simply averaging or taking the median might not be insightful. What would be the best way to aggregate or transform this data?

Thank you!

[dattatreyam23]

Hi @ramadhita-ind,

Thank you for reaching out to the Google Meridian support team.

Since you have a 4-hour granularity for your weather data, we recommend you to aggregate your weather data to match the granularity of the rest of your data, and include it as a control variable. You could use the mean or median temperature but could include additional measures as controls such as the min or max temperature, pressure, rainfall, etc depending on what makes sense for your use-case.

It would be beneficial to understand the granularity of your KPI data, since having a finer granularity for your weather data wouldn’t yield any additional advantage. Weekly granularity is a general recommendation but depending on your use case, it can vary - e.g., for a use case where marketing strategies are updated daily, finer granularity makes more sense.

Feel free to reach out if you have any other questions or suggestions regarding Meridian.

Thank you,

Google Meridian Support Team