Insights are calculated every day after midnight, taking into account the local time of the place where the product is installed. Local time depends on the time zone configured on the location or thing.
Input Data
During the computation, the underlying insight algorithm is supplied with inputs, for instance:
the context objects (thing, location, customer)
for each referenced metric, the values extracted based on the specified data range and limited to the end (local time) of the day before the time of computation.
the events (failures, anomalies, work-sessions) occurred during the day before the time of computation (only for event-based predefined insights).
Data Granularity
For predefined insight metrics, a daily sample is generated by using the timestamp at the beginning of the previous day (local time) before the time the computation takes place.
Instead, for custom insight metrics, a sample is saved for each sampling period (minimum 1 day). Even if the sampling period is longer than one day (e.g., one month), insight is calculated daily, and the current value is updated. Each value is saved by using the timestamp at the beginning of the sampling period (local time).
The calculation starts from when the insight metric is configured, or the product is registered in the DPS; historical data prior to that time will not be available.
Insight Properties
In the presence of metrics based on predefined insights, once the insight computation is completed, context objects will be updated with aggregated values. For more details refer to the Predefined Insight Properties article.