PostgreSQL

SQL is where most data already lives, and PostgreSQL brings a lot of statistics with it: avg, stddev_samp, var_samp, percentile_cont, corr, and regr_slope are all built-in aggregates. The from-scratch versions avoid them and use only basic aggregates such as sum, count, min, and max, plus ordinary arithmetic and window functions. SQL describes results rather than steps, so iterative methods take some ingenuity: k-means repeats its two steps with a recursive query, WITH RECURSIVE.

Quirks

• The runner loads data.csv into a table named data, with numeric columns as double precision and a row_id column numbering rows in file order, from 1. Tables otherwise have no order.
• The final query returns two columns, name and value, one row per result.
• Integer division truncates: 7 / 2 is 3, and 7 / 2.0 is 3.5.

git clone https://github.com/qnixon/dswithx.git
cd dswithx
python3 harness/run_tests.py --language postgresql