Data Analysis

Data Analysis (Multidimensional Analysis) aggregates many recorded invocations so you can tell a one-off spike from a systemic regression. Open it from Analytics → Data Analysis. You must set a date range and an application. Optional conditions match the Data Browser form. You can also apply a saved chart when every rule on that chart selects the same application.

Screenshot placeholder: Workstation → Analytics → Data Analysis. Percentile curve, heatmap, and Exceptions panel after a search; AI Analysis in the toolbar.

The view has three areas. After a search, AI Analysis (if enabled) opens the AI Assistant grounded in this query.

Duration distribution

• A percentile curve of method duration with bars for invocation count in each percentile band. The count in a band is the difference between consecutive percentiles (for example P50 minus P25), not a standard histogram bucket.

• A table of exact percentile values, durations, and counts.

• Percentiles over time, and invocation count over time.

2D analysis

A heatmap of a selected data type against duration (vertical) and time (horizontal). Cells aggregate the range they cover:

• Count — invocations in the cell

• Off-CPU — share of invocations whose off-CPU time is largest

• Exceptions — methods that threw

A percentile control limits the chart to invocations below that percentile of duration.

Exceptions

• Top root exceptions — the most frequent originating exceptions.

• Selecting one fills Top method exception — methods that throw it.

• The stack-trace control opens successive traces for that exception.

From here you typically open an individual call in the Method Execution Tree.