online-ml/watermill.rs

🦀 Online statistics in Rust

Online statistics in Rust 🦀

watermill is crate 🦀 for Blazingly fast, generic and serializable online statistics.

Quickstart

---

Let's compute the online median and then serialize it:

use watermill::quantile::Quantile;
use watermill::stats::Univariate;
let data: Vec<f64> = vec![9., 7., 3., 2., 6., 1., 8., 5., 4.];
let mut running_median: Quantile<f64> = Quantile::new(0.5_f64).unwrap();
for x in data.into_iter() {
    running_median.update(x); // update the current statistics
    println!("The actual median value is: {}", running_median.get());
}
assert_eq!(running_median.get(), 5.0);

// Convert the statistic to a JSON string.
let serialized = serde_json::to_string(&running_median).unwrap();

// Convert the JSON string back to a statistic.
let deserialized: Quantile<f64> = serde_json::from_str(&serialized).unwrap();

Now let's compute the online sum using the iterators:

use watermill::iter::IterStatisticsExtend;
let data: Vec<f64> = vec![1., 2., 3.];
let vec_true: Vec<f64> = vec![1., 3., 6.];
for (d, t) in data.into_iter().online_sum().zip(vec_true.into_iter()) {
    assert_eq!(d, t); //       ^^^^^^^^^^
}

You can also compute rolling statistics; in the following example let's compute the rolling sum on 2 previous data:

use watermill::rolling::Rolling;
use watermill::stats::Univariate;
use watermill::variance::Variance;
let data: Vec<f64> = vec![9., 7., 3., 2., 6., 1., 8., 5., 4.];
let mut running_var: Variance<f64> = Variance::default();
// We wrap `running_var` inside the `Rolling` struct.
let mut rolling_var: Rolling<f64> = Rolling::new(&mut running_var, 2).unwrap();
for x in data.into_iter() {
    rolling_var.update(x);
}
assert_eq!(rolling_var.get(), 0.5);

Installation

---

Add the following line to your cargo.toml:

[dependencies]
watermill = "0.1.0"

Statistics available

Statistics	Rollable ?
Mean	✅
Variance	✅
Sum	✅
Min	✅
Max	✅
Count	❌
Quantile	✅
Peak to peak	✅
Exponentially weighted mean	❌
Exponentially weighted variance	❌
Interquartile range	✅
Kurtosis	❌
Skewness	❌
Covariance	❌

Inspiration

---

The stats module of the river library in Python greatly inspired this crate.