Functional Select With Aggregations

Question

Given the following table:

t:([] sym:`AAPL`MSFT`GOOG`TSLA`NVDA; b:10 20 15 30 25)

Write the functional equivalent of:

select minB:min b, avgB:avg b from t

Expected result:

minB avgB
---------
10   20

Solution

?[t; (); 0b; `minB`avgB!((min;`b);(avg;`b))]

Explanation

The query computes two aggregations across the whole table — no grouping, no filtering. Mapping that to the four-argument functional form:

When the output column is a computation rather than a plain pass-through, the dictionary value is a small list — a parse tree — describing the operation: (function; column).

(min;`b)    / apply min to column b → minB
(avg;`b)    / apply avg to column b → avgB

The full select dictionary:

`minB`avgB!((min;`b);(avg;`b))
/ minB -> min b
/ avgB -> avg b

Verifying with parse

If you are ever unsure of the functional form for a particular expression, parse will reveal it. Pass your qSQL as a string and read back the structure q would use internally:

parse "select minB:min b, avgB:avg b from t"

?
`t
()
0b
`minB`avgB!((min;`b);(avg;`b))

Each line is one of the four arguments, in order. The last line matches the dictionary in the solution exactly.

A few things to watch when reading parse output: it uses q's underlying k notation, so built-in functions may appear with unfamiliar names, and symbol values in conditions will have a leading , (k's way of writing enlist) that you need to replace when writing the functional form yourself.