I use DuckDB because I can express complex analytic queries better using a SQL mental model. Most software people hate SQL because they can never remember its syntax, but for data scientists, SQL lets us express our thoughts more simply and precisely in a declarative fashion — which gets us query plan optimization as a plus.
People have been trying to get rid of SQL for years yet they only end up reinventing it badly.
I’ve written a lot of code and the two notations I always gravitate toward are the magrittr + dplyr pipeline notation and SQL.
The chained methods notation is a bit too unergonomic especially to express window functions and complex joins.
Spark started out with method chaining but eventually found that most people used Spark SQL.
People have been trying to get rid of SQL for years yet they only end up reinventing it badly.
I’ve written a lot of code and the two notations I always gravitate toward are the magrittr + dplyr pipeline notation and SQL.
The chained methods notation is a bit too unergonomic especially to express window functions and complex joins.
Spark started out with method chaining but eventually found that most people used Spark SQL.