MIM Overview | How does my model determine trends?

With our latest version of the model, we have improved trend detection to account for variation in your data. How do we do this? Great question! 

We take the first and last dates of your data and divide it into chunks. Trends are then detected within that chunk. 

For example, let's say you data begins January 1, 2022. You started following your model recommendations January 2023. At this point, your baseline data probably varies from a year prior, so going forward, you want your model to be smarter and learn from this new behavior and not old trends.

So on January 1, 2024, your model will take your two years (730 days) work of data and spilt it into, for example, ten chunks* of 73 days. In conjunction with seasonality and other trends, your model will be making recommendations based on your more recent spent habits. 

*Number of chunks may vary.