wbi-rs 0.1.11

A Rust library + CLI to fetch, store, visualize, and summarize World Bank indicator data.
Documentation 
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
51
52
53
54
55
56
57
58
59
60
61

//! LOESS (locally weighted linear regression) smoother implementation.

/// Simple LOESS (locally weighted linear regression) smoother for yearly series.
/// `span` is the fraction of neighbors used (0 < span <= 1).
pub fn loess_series(xs: &[f64], ys: &[f64], span: f64) -> Vec<f64> {
    let n = xs.len();
    if n == 0 {
        return vec![];
    }
    let span = span.clamp(1.0 / n as f64, 1.0);
    let window = ((n as f64 * span).ceil() as usize).max(2);
    let mut yhat = vec![0.0; n];
    for i in 0..n {
        // Find window of nearest neighbors around i
        let mut idx: Vec<usize> = (0..n).collect();
        idx.sort_by(|&a, &b| {
            (xs[a] - xs[i])
                .abs()
                .partial_cmp(&(xs[b] - xs[i]).abs())
                .unwrap()
        });
        let idxw = &idx[..window];
        let max_d = (xs[*idxw.last().unwrap()] - xs[i]).abs();
        // Weights: tricube kernel
        // Standard mathematical notation for weighted least squares
        #[allow(clippy::similar_names)]
        let mut sw = 0.0;
        #[allow(clippy::similar_names)]
        let mut swx = 0.0;
        #[allow(clippy::similar_names)]
        let mut swy = 0.0;
        #[allow(clippy::similar_names)]
        let mut swxx = 0.0;
        #[allow(clippy::similar_names)]
        let mut swxy = 0.0;
        for &j in idxw {
            let d = (xs[j] - xs[i]).abs();
            let u = if max_d == 0.0 {
                0.0
            } else {
                (d / max_d).min(1.0)
            };
            let w = (1.0 - u * u * u).powi(3);
            sw += w;
            swx += w * xs[j];
            swy += w * ys[j];
            swxx += w * xs[j] * xs[j];
            swxy += w * xs[j] * ys[j];
        }
        // Weighted linear regression y = a + b x
        let denom = sw * swxx - swx * swx;
        if denom.abs() < 1e-12 {
            yhat[i] = swy / sw.max(1e-12);
        } else {
            let b = (sw * swxy - swx * swy) / denom;
            let a = (swy - b * swx) / sw;
            yhat[i] = a + b * xs[i];
        }
    }
    yhat
}