Skip to main content

theta/
theta.rs

1//! Demonstrates the Theta sketch family: cardinality estimation, set
2//! operations (union, intersection, a-not-b), and Jaccard similarity.
3//!
4//! Run with:
5//!   cargo run --example theta --features theta
6
7use apache_datasketches::theta::{
8    jaccard_similarity, ThetaAnotB, ThetaIntersection, ThetaSketchBuilder, ThetaUnionBuilder,
9};
10
11fn main() {
12    // Build two sketches representing two overlapping sets of user IDs.
13    let mut visitors_day1 = ThetaSketchBuilder::new().lg_k(12).build().unwrap();
14    for id in 0..10_000u64 {
15        visitors_day1.update_u64(id);
16    }
17
18    let mut visitors_day2 = ThetaSketchBuilder::new().lg_k(12).build().unwrap();
19    for id in 5_000..15_000u64 {
20        visitors_day2.update_u64(id);
21    }
22
23    println!(
24        "Day 1 unique visitors (estimate): {:.0}",
25        visitors_day1.get_estimate()
26    );
27    println!(
28        "Day 2 unique visitors (estimate): {:.0}",
29        visitors_day2.get_estimate()
30    );
31
32    // Union: total unique visitors across both days.
33    let mut union = ThetaUnionBuilder::new().lg_k(12).build().unwrap();
34    union.update(&visitors_day1);
35    union.update(&visitors_day2);
36    let total_unique = union.get_result(true);
37    println!(
38        "Total unique visitors (union estimate): {:.0}",
39        total_unique.get_estimate()
40    );
41
42    // Intersection: visitors who came back on day 2.
43    let mut intersection = ThetaIntersection::new();
44    intersection.update(&visitors_day1);
45    intersection.update(&visitors_day2);
46    match intersection.get_result(true) {
47        Ok(returning) => println!(
48            "Returning visitors (intersection estimate): {:.0}",
49            returning.get_estimate()
50        ),
51        Err(e) => println!("No intersection result: {e}"),
52    }
53
54    // A-not-b: visitors who only came on day 1.
55    let a_not_b = ThetaAnotB::new();
56    let day1_only = a_not_b.compute(&visitors_day1, &visitors_day2, true);
57    println!(
58        "Day-1-only visitors (a-not-b estimate): {:.0}",
59        day1_only.get_estimate()
60    );
61
62    // Jaccard similarity: how similar are the two days' visitor sets?
63    let similarity = jaccard_similarity(&visitors_day1, &visitors_day2);
64    println!(
65        "Jaccard similarity: {:.3} (range [{:.3}, {:.3}])",
66        similarity.estimate, similarity.lower_bound, similarity.upper_bound
67    );
68
69    // Serialize a compact sketch for storage/transmission, then restore it.
70    let compact = visitors_day1.compact(true);
71    let bytes = compact.serialize_compact();
72    println!("Serialized day-1 sketch: {} bytes", bytes.len());
73    let restored = apache_datasketches::theta::CompactThetaSketch::deserialize(&bytes).unwrap();
74    println!("Restored estimate: {:.0}", restored.get_estimate());
75}