1.2.3 Lazy API🔗ℹ

Eager verbs run immediately on a dataframe. The same verbs applied to a lazyframe (from lazy or scan-csv) build a query plan that collect optimises and runs. Here iris-csv is a path to a small CSV.

> (define df-small (~> (read-csv iris-csv) (filter (> (col "sepal_length") 5))))

> (~> df-small (group-by "species") (agg (mean "sepal_width")))

shape: (3, 2) ┌────────────┬─────────────┐ │ species    ┆ sepal_width │ │ ---        ┆ ---         │ │ str        ┆ f64         │ ╞════════════╪═════════════╡ │ virginica  ┆ 2.975       │ │ setosa     ┆ 3.7         │ │ versicolor ┆ 2.95        │ └────────────┴─────────────┘

> (define q     (~> (scan-csv iris-csv)         (filter (> (col "sepal_length") 5))         (group-by "species")         (agg (mean "sepal_width"))))

> (collect q)

shape: (3, 2) ┌────────────┬─────────────┐ │ species    ┆ sepal_width │ │ ---        ┆ ---         │ │ str        ┆ f64         │ ╞════════════╪═════════════╡ │ versicolor ┆ 2.95        │ │ setosa     ┆ 3.7         │ │ virginica  ┆ 2.975       │ └────────────┴─────────────┘

explain previews the plan collect would run, after optimisation: the filter has moved into the CSV scan.

> (displayln (explain q))

AGGREGATE[maintain_order: false]   [col("sepal_width").mean()] BY [col("species")]   FROM   simple π 2/2 ["species", "sepal_width"]     Csv SCAN [/home/root/user/.local/share/racket/9.3/pkgs/polars/scribblings/guide/iris.csv]     PROJECT 3/5 COLUMNS     SELECTION: col("sepal_length") > 5.0     ESTIMATED ROWS: 19

API gap: no schema-only LazyFrame.