Discovery Settings

In the Discover settings tab, you will find all configurations regarding the event and the property catalog creation.


  • Lookback Days -> Mitzu looks at the data for certain patterns during the event catalog creation. This analysis is done purely in SQL on a small sample of the raw events. This option narrows down the sampling to the last X days. Where X is the Lookback Days value. In practice, this means some events that no longer appear in multi-event tables won't appear in the event catalog. Another implication is that columns that don't have values in the last X days won't appear in the event property catalog. Collecting possible filter values is also limited to the last X days.

  • Sample Size -> the sample size to find filter values for each event property. Events that don't appear in the sample from multi-event tables won't be listed in the event catalog.

  • Max Enum Cardinality -> The maximum number of filter values collected from the table columns. In case of a finite cardinality event property (country code, locale, platform) Mitzu will collect the values up to this number. If the possible values for an event property exceed this number, Mitzu won't collect any values (user_id, event_id, etc).

  • Iterative Discovery Column Chunk Size -> Some data models have tables with 1000 columns stored in large tables. This setting will allow Mitzu to collect event properties from these tables in chunks.

  • User Table Sample Size -> User tables are scanned for possible user properties and values. In this case, this sample size config will apply to sample the user table.

Last updated