Skip to content

Glossary

Quick Lookup

Common terms used in this module:

  • Baseline - a reference period representing expected or normal system behavior
  • Reference Dataset - the dataset representing baseline system behavior
  • Current Dataset - the dataset representing recent system observations
  • Baseline Comparison - comparing current metrics to a reference baseline
  • Change Detection - identifying meaningful differences between time periods
  • Statistical Drift Detection - using statistical methods to determine whether changes are significant
  • Drift - a sustained change in system behavior over time, observed when recent data consistently differs from a historical reference or baseline.

Baseline

A baseline represents expected system behavior based on a historical period. Analysts often compare current data to a baseline to determine whether the system is behaving normally.

Reference Dataset

A dataset representing the baseline period used for comparison. Example: last week's system metrics.

Current Dataset

A dataset representing recent system behavior. Example: today's or this week's system metrics.

Baseline Comparison

The process of comparing summary statistics between two time periods. Baseline comparisons help analysts identify potential changes in system behavior. Example:

reference_avg_latency
current_avg_latency
difference = current - reference

Change Detection

The process of identifying meaningful differences between datasets or time periods.

Drift

Drift occurs when system behavior changes persistently over time relative to a historical baseline. Baseline comparisons can provide early evidence that drift may be occurring. Examples include:

  • increasing latency
  • rising error rates
  • declining throughput

Statistical Drift Detection

A more advanced approach that uses statistical tests to determine whether observed changes are likely due to random variation or represent a true shift in system behavior. These techniques provide stronger evidence that system behavior has changed. Examples include:

  • distribution comparison tests
  • hypothesis tests
  • statistical control methods

Summary Statistics

Simple numerical summaries used to describe a dataset. Summary statistics help compare system behavior between two periods. Common examples include:

  • mean (average)
  • minimum
  • maximum
  • count