Aggregation
A grid tells you which records. Aggregation tells you how many, how much and which is highest — the shape most questions actually have.
Open the aggregation panel from the grid you are working in. It runs against the same query, so everything you have filtered still applies.
The two decisions
Group by — the field to break the data down along: revenue centre, employee, detector, hour, location.
Measure — what to work out for each group:
| Measure | Answers |
|---|---|
| Count | How many records |
| Total | The sum of a numeric field |
| Average | The mean per record |
| Min / Max | The smallest and largest values |
| First / Last | The earliest and latest |
| Unique … | How many distinct values a column had, rather than how many rows |
"Voids by register" is group by register, measure count. "Average check by revenue centre" is group by revenue centre, measure average on the check amount.
Unique is the one people miss, and it answers a different question. Counting detections tells you how busy a detector was; counting Unique Subject tells you how many different people that was — one person seen forty times and forty people seen once are the same count and very different findings. The same distinction applies to unique plates, unique patrons, unique tickets and unique shoes, wherever the underlying data carries them.
Which numeric fields you can total or average depends on the view you are aggregating: subject confidence, demographics age, visit duration, bet session figures and the like are offered where they exist.
Choosing a grouping field
The grouping field decides whether the result is useful. Group by something with a manageable number of distinct values — a dozen registers is readable, ten thousand check numbers is not.
If a grouping produces hundreds of rows, group by something coarser and drill down afterwards. That is faster than reading a long list, and it is the pattern the drill-down is built for.
Grouping by duration
Where a view records how long something lasted — a zone visit, an observed presence — duration can be grouped in bands rather than by its exact number of seconds. That is what turns "every visit with its own row" into "how many visits were under five minutes, five to fifteen, and longer".
It is the natural grouping for dwell time, and the one to reach for before exporting a long list and banding it in a spreadsheet.
Grouping by time
Time is the most useful grouping field in eConnect, because almost everything is a pattern over time — visits by hour, transactions by day, detections by week.
Choose a granularity that matches the period: by hour over a day, by day over a month, by week over a quarter. Too fine and the shape disappears into noise; too coarse and it disappears into a single bar.
Presets
Common aggregations are available as presets, which is worth checking before building one by hand — the question you are asking may already be there.
Reading the result
The panel shows the groups and their measures. Sort by the measure to put the largest first, which is usually the point: you are looking for the outlier.
Then look at the chart
Numbers show precision; a chart shows shape. Switch to the chart view and the outlier is usually obvious at a glance.
More importantly, a chart is clickable — selecting a slice takes you to the rows behind it. Going from a total to its records in one click is the most useful movement in eConnect.
Exporting
Aggregations export their totals rather than the underlying rows. See Exporting Data.
Audit record
What eConnect writes to the audit trail for the actions on this page.
QUERY EXPORT '{name}' FROM {start date} to {end date}