liveBook · Manning

Snapshot

As explained in Machine Learning Platform Engineering

A data snapshot is a persisted representation of data captured at a particular point in time.

A data snapshot is a persisted representation of data captured at a particular point in time. In monitoring workflows, snapshots provide the historical records needed to compare data across intervals and visualize changes in a dashboard. section 6.2.1

Use snapshots to organize monitoring history:

  • Store snapshots inside projects, with a workspace acting as the directory that organizes projects and their logs. section 6.2.1
  • Compare current and reference datasets to produce data-drift reports and track metrics over time. section 6.2.1
  • Select snapshots through dashboard filters and render their measurements as counters, plots, bars, scatters, or histograms. section 6.2.1

Organizing income-classifier snapshots

A workspace can contain an income-classifier project with files such as snapshot_1.json and snapshot_2.json. A RemoteWorkspace connected to the local Evidently UI creates the project, after which dashboard panels can display drift measurements across its snapshots.

How Machine Learning Platform Engineering covers it

How deep it goes
Explains storage in workspace projects and dashboard filtering, but defers snapshot-generation details.
Examples
Uses JSON snapshot files organized under projects and an income-classifier drift dashboard.
What you can do after
Operate/configure Evidently workspaces and dashboards to organize snapshots and compare monitored data over time.
How it connects
Connects data snapshots with data drift detection, reference data, current data, workspaces, projects, and dashboards.

Also explained in

Compare how 11 books explain snapshot →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Machine Learning Platform Engineering. The section links point to the parts of the book it draws on.