Columns

[Polars Documentation]

pl.col("A") or pl.col.A starts an expression by selecting a column named “A”. While the Polars Library allows for multiple columns to be selected simultaneously (via pl.col("*"), pl.col("A", "B"), pl.col(pl.String), pl.exclude, and so on), the OpenDP Library currently only supports selection of one column at a time. The column name may be changed via .alias.

Take for example the work hours dataset, where there are a collection of columns labeled METHODX, where X is an increasing alphabetic sequence.

import polars as pl
import opendp.prelude as dp

dp.enable_features("contrib")

# not recommended, OpenDP will reject this joint expression over multiple columns
single_expr = pl.col([f"METHOD{l}" for l in "ABCDE"]).fill_null(0).dp.sum((0, 9))

# build individual expressions for each query
split_exprs = [pl.col(f"METHOD{l}").fill_null(0).dp.sum((0, 9)) for l in "ABCDE"]

Demonstration of use:

context = dp.Context.compositor(
    data=pl.scan_csv(dp.examples.get_france_lfs_path(), ignore_errors=True),
    privacy_unit=dp.unit_of(contributions=36),
    privacy_loss=dp.loss_of(epsilon=1.0),
    split_evenly_over=1,
    margins=[
        dp.polars.Margin(max_partition_length=60_000_000 * 36)
    ],
)

context.query().select(split_exprs).release().collect()

shape: (1, 5)

METHODA	METHODB	METHODC	METHODD	METHODE
i64	i64	i64	i64	i64
1704484	1699390	1702886	1703232	1705356