The following is a brief description of how Chainlink Oracle Reputation gathers its data
To sort for isolated metrics, the COR engine monitors each individual Chainlink related log and transaction request. The logs and transactions are aggregated into snapshots. Snapshots are collected on a per-oracle basis and come in two forms, instantaneous snapshots and cumulative snapshots.
State snapshots of the network every ten minutes on a per-oracle basis. All transactions that an oracle is connected to within this period are added to this snapshot, from which all metrics are calculated.
Cumulative snapshots are the aggregate of all of the metrics we collect on an oracle. The way this works is that first the instantaneous snapshot is calculated, and then its data merged with the latest cumulative snapshot.
Network snapshots are only used in the Chainlink network dashboard. These snapshtos represent a summaation of data over the last 7 days from any given point in time. This allows us to analyse the Chainlink network on a rolling 7 day basis, seeing its average performance over the last 7 days at any given point in time.