Shrink numeric columns to the minimal required datatype

Source code

Description

Shrink to the dtype needed to fit the extrema of this Series. This can be used to reduce memory pressure.

Usage

<Expr>$shrink_dtype()

Value

Expr

Examples

library("polars")

df = pl$DataFrame(
  a = 1:3,
  b = c(1, 2, 3)
)
df

#> shape: (3, 2)
#> ┌─────┬─────┐
#> │ a   ┆ b   │
#> │ --- ┆ --- │
#> │ i32 ┆ f64 │
#> ╞═════╪═════╡
#> │ 1   ┆ 1.0 │
#> │ 2   ┆ 2.0 │
#> │ 3   ┆ 3.0 │
#> └─────┴─────┘

df$with_columns(pl$all()$shrink_dtype()$name$suffix("_shrunk"))

#> shape: (3, 4)
#> ┌─────┬─────┬──────────┬──────────┐
#> │ a   ┆ b   ┆ a_shrunk ┆ b_shrunk │
#> │ --- ┆ --- ┆ ---      ┆ ---      │
#> │ i32 ┆ f64 ┆ i8       ┆ f32      │
#> ╞═════╪═════╪══════════╪══════════╡
#> │ 1   ┆ 1.0 ┆ 1        ┆ 1.0      │
#> │ 2   ┆ 2.0 ┆ 2        ┆ 2.0      │
#> │ 3   ┆ 3.0 ┆ 3        ┆ 3.0      │
#> └─────┴─────┴──────────┴──────────┘