Metric Type Description
1 Metric Type Definitions
Guandata Metrics Center currently supports creating the following metric types:
| Metric Classification | Description |
|---|---|
| Atomic Metrics | Based on measurements under a specific business event behavior, these are metrics that cannot be further decomposed in business definitions, such as net profit = sum(order net profit), total transaction volume = count(distinct order number). |
| Composite Metrics | Calculated from multiple metrics (atomic metrics, composite metrics) through arithmetic operations (addition, subtraction, multiplication, division), such as Channel A sales ratio = Channel A sales / (Channel A sales + Channel B sales) or Channel A sales / Total sales. |
| Derived Metrics | Derived metrics are created based on a single atomic or composite metric, combined with cumulative calculations, year-over-year comparisons, and other derivation methods to extend analysis, such as creating year-over-year net profit based on net profit, or cumulative values for the past N days. |
The relationship between the three types for mutual creation is shown in the figure below:
2 Example Description
Taking a retail scenario as an example, the following is a retail transaction table. You can create metrics about transaction performance based on fields such as order ID, order quantity, order net profit, etc. in the table, and use order date, store region, store city, etc. as analysis dimensions for the metrics.
For detailed processes of creating atomic metrics, composite metrics, and derived metrics, see: