opendp 0.14.2-dev.20260401.2

A library of differential privacy algorithms for the statistical analysis of sensitive private data.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50

use polars::{
    df,
    prelude::{IntoLazy, NamedFrom, col},
};

use crate::{
    domains::{AtomDomain, LazyFrameDomain, Margin, SeriesDomain},
    measurements::make_private_lazyframe,
    measures::MaxDivergence,
    metrics::SymmetricDistance,
};

use super::*;

#[test]
fn test_dp_quantile() -> Fallible<()> {
    let lf_domain = LazyFrameDomain::new(vec![SeriesDomain::new(
        "A",
        AtomDomain::<f64>::new_non_nan(),
    )])?
    .with_margin(Margin::select().with_max_length(100))?;

    let lf = df!("A" => [50f64; 100])?.lazy();

    let candidates = Series::new(
        "".into(),
        [0., 10., 20., 30., 40., 50., 60., 70., 80., 90., 100.],
    );

    let m_quant = make_private_lazyframe(
        lf_domain,
        SymmetricDistance,
        MaxDivergence,
        lf.clone().select([col("A").dp().median(candidates, None)]),
        Some(1.),
        None,
    )?;

    let observed = m_quant
        .invoke(&lf)?
        .collect()?
        .column("A")?
        .f64()?
        .get(0)
        .unwrap();

    let expected = 50.;
    assert_eq!(observed, expected);
    Ok(())
}