# Analytics queries

A console program that uses `TAtlasAnalytics` directly, without a report
definition or a dashboard: it projects and filters a `TAtlasTable` with
`Select`, groups it with `Aggregate`, shows how missing values and the source
table's own filter affect the results, catches the two exception classes and
exports the version 1 dataset JSON that the web example reads.

## What it demonstrates

- `TAtlasAnalytics.Select` with a field list and typed, conjunctive filters
  (`afGreater` on a numeric field, `afContains` on a text field, case-insensitive
  equality).
- `TAtlasAnalytics.Aggregate` with two dimensions, and ungrouped with an
  average that skips a missing value while the count still includes the row.
- `TAtlasAnalytics.FieldIndex` to find a column by name without regard to case.
- Queries reading the source's current view (`TAtlasTable.SetFilter`) and never
  modifying it.
- `EAtlasAnalyticsError` for an unknown field and `EAtlasDataError` for a filter
  value that does not fit the column kind.
- `TAtlasAnalytics.DatasetJSON` output: numbers as JSON numbers, missing values
  as `null`.

## Build and run

Open `AnalyticsQueries.dpr` in Delphi 13; the IDE creates the project file. Add the Atlas Analytics `src` folder and the Atlas VCL `src` folder to the project's search path, then build and run for Win32 or Win64. No packages need to be installed. The program needs no input. It checks its own results and stops with an error if any of them differs from the output below.

## Expected output

```
Sales above 10000: 3 rows, 2 columns
Desktop sales in regions containing "e": 2 rows
    East  Desktop  18000 1
    West  Desktop  22500 1
    East  Service   6400 1
   North  Desktop  12000 1
    West  Service   9600 1
   North  Service   4800 1
Average revenue 12216.6666666667 over 7 rows
Rejected: Unknown field: Margin
Rejected: Invalid finite number: lots
Dataset JSON: {"version":1,"fields":[{"name":"Region","kind":"text"},{"name":"Product","kind":"text"},{"name":"Revenue","kind":"number"},{"name":"Units","kind":"number"}],"rows":[["East","Desktop",18000.0,12.0],["West","Desktop",22500.0,15.0],["East","Service",6400.0,8.0],["North","Desktop",12000.0,8.0],["West","Service",9600.0,12.0],["North","Service",4800.0,6.0],["East","Service",null,1.0]]}
PASS
```

Numbers are written as JSON numbers with a decimal point (`18000.0`), and the
empty revenue cell of the last row is `null`. The exit code is 0; any mismatch
prints `Check failed: ...` and exits with 1.

## Adapting it

Replace the `AddColumn`/`AddRow` calls with your own table (or `ImportCSV` of a
file) and change the field names in the filters and measures. Every measure
except `amCount` needs an `ackNumber` column.