Changepoint detection with Pruned Exact Linear Time.

Usage

Python

from pelt import predict

predict(signal, penalty=20, segment_cost_function="l1", jump=10, minimum_segment_length=2)

Rust

use std::num::NonZero;
use pelt::{Pelt, SegmentCostFunction};

// Setup the structure for calculating changepoints
let pelt = Pelt::new()
  .with_jump(NonZero::new(5).expect("Invalid number"))
  .with_minimum_segment_length(NonZero::new(2).expect("Invalid number"))
  .with_segment_cost_function(SegmentCostFunction::L1);

// Do the calculation on a data set
let penalty = 10.0;
let result = pelt.predict(&signal[..], penalty)?;

Run locally

# Install maturin inside a Python environment
python3 -m venv .env
source .env/bin/activate
pip install maturin numpy

# Create a Python package from the Rust code
maturin develop

# Open an interpreter
python

>>> from pelt import predict
>>> import numpy as np
>>> signal = np.array([np.sin(np.arange(0, 1000, 10))]).transpose()
>>> predict(signal, penalty=20)

Benchmarks

Warning

Like all benchmarks, take these with a grain of salt.

Python

Comparison with ruptures:

Cost Function	Data Points	Data Dimension	Mean pelt	Mean ruptures	Times Faster
L2	100	1D	2.065 μs	2.820 ms	1365.7x
L2	100	2D	2.294 μs	2.817 ms	1228.1x
L2	1000	1D	107.256 μs	171.377 ms	1597.8x
L2	1000	2D	51.096 μs	90.582 ms	1772.8x
L2	10000	1D	20.038 ms	11.454 s	571.6x
L2	10000	2D	2.751 ms	1.672 s	607.7x
L1	100	1D	9.148 μs	4.312 ms	471.3x
L1	100	2D	19.886 μs	4.774 ms	240.1x
L1	1000	1D	173.917 μs	162.959 ms	937.0x
L1	1000	2D	2.269 ms	588.634 ms	259.5x
L1	10000	1D	3.515 ms	14.674 s	4174.3x
L1	10000	2D	104.050 ms	30.726 s	295.3x

Command

maturin develop
python benches/bench_compare.py

Profile

Command

cargo build --example simple --profile profiling \
 && samply record target/profiling/examples/simple tests/signals-large.csv

Credits

• fastpelt
• ruptures