golang library for computing matrix profiles along with other time series analysis features
Golang library for computing a matrix profiles and matrix profile indexes. Features also include time series discords, time series segmentation, and motif discovery after computing the matrix profile. Visit The UCR Matrix Profile Page for more details into matrix profiles.
$ go get github.com/matrix-profile-foundation/go-matrixprofile
// example_mp.go
package main
import (
"fmt"
mp "github.com/matrix-profile-foundation/go-matrixprofile"
)
func main() {
sig := []float64{0, 0.99, 1, 0, 0, 0.98, 1, 0, 0, 0.96, 1, 0}
p, err := mp.New(sig, nil, 4)
if err != nil {
panic(err)
}
if err = p.Compute(nil); err != nil {
panic(err)
}
fmt.Printf("Signal: %.3f\n", sig)
fmt.Printf("Matrix Profile: %.3f\n", p.MP)
fmt.Printf("Profile Index: %5d\n", p.Idx)
}
$ go run example_mp.go
Signal: [0.000 0.990 1.000 0.000 0.000 0.980 1.000 0.000 0.000 0.960 1.000 0.000]
Matrix Profile: [0.014 0.014 0.029 0.029 0.014 0.014 0.029 0.029 0.029]
Profile Index: [ 4 5 6 7 0 1 2 3 4]
Going through a completely synthetic scenario, we'll cover what features to look for in a matrix profile, and what the additional Discords, TopKMotifs, and Segment tell us. We'll first be generating a fake signal that is composed of sine waves, noise, and sawtooth waves. We then run STOMP on the signal to calculte the matrix profile and matrix profile indexes.
subsequence length: 32
The code to generate the graph can be found in this example.
Based on [4] we can extend the matrix profile algorithm to multi-dimensional scenario.
subsequence length: 25
The plots can be generated by running
$ make example
go test ./... -run=Example
ok github.com/matrix-profile-foundation/go-matrixprofile 0.256s
ok github.com/matrix-profile-foundation/go-matrixprofile/av (cached) [no tests to run]
ok github.com/matrix-profile-foundation/go-matrixprofile/siggen (cached) [no tests to run]
ok github.com/matrix-profile-foundation/go-matrixprofile/util (cached) [no tests to run]
A png file will be saved in the top level directory of the repository as mp_sine.png
and mp_kdim.png
BenchmarkMStomp-4 39 29853485 ns/op 7336245 B/op 227071 allocs/op
BenchmarkZNormalize-4 7112282 185 ns/op 256 B/op 1 allocs/op
BenchmarkMovmeanstd-4 89810 13628 ns/op 32768 B/op 4 allocs/op
BenchmarkCrossCorrelate-4 15190 75262 ns/op 24584 B/op 3 allocs/op
BenchmarkMass-4 15421 78660 ns/op 24842 B/op 4 allocs/op
BenchmarkDistanceProfile-4 15190 79092 ns/op 24842 B/op 4 allocs/op
BenchmarkCalculateDistanceProfile-4 220363 4625 ns/op 0 B/op 0 allocs/op
BenchmarkStmp/m32_pts1k-4 15 77806814 ns/op 24209736 B/op 3892 allocs/op
BenchmarkStmp/m128_pts1k-4 16 70673766 ns/op 22496294 B/op 3508 allocs/op
BenchmarkStamp/m32_p2_pts1k-4 25 46207243 ns/op 24284148 B/op 3909 allocs/op
BenchmarkStomp/m128_p1_pts__1024-4 152 7740858 ns/op 196805 B/op 28 allocs/op
BenchmarkStomp/m128_p2_pts__4096-4 13 81826774 ns/op 1116937 B/op 39 allocs/op
BenchmarkStomp/m128_p2_pts_16384-4 1 1342203283 ns/op 4776832 B/op 45 allocs/op
BenchmarkStomp/m128_p4_pts_16384-4 1 1269550826 ns/op 7153728 B/op 67 allocs/op
BenchmarkStomp/m1024_p2_pts_16384-4 1 1235325258 ns/op 4776832 B/op 45 allocs/op
BenchmarkMpx/m128_p1_pts__1024-4 564 2310017 ns/op 84591 B/op 26 allocs/op
BenchmarkMpx/m128_p2_pts__4096-4 63 19927988 ns/op 400206 B/op 32 allocs/op
BenchmarkMpx/m128_p2_pts_16384-4 4 294076163 ns/op 1708912 B/op 33 allocs/op
BenchmarkMpx/m128_p4_pts_16384-4 4 327366290 ns/op 2237776 B/op 45 allocs/op
BenchmarkMpx/m1024_p2_pts_16384-4 4 330582811 ns/op 1737584 B/op 33 allocs/op
BenchmarkUpdate-4 80 13849082 ns/op 795065 B/op 18 allocs/op
Ran on a 2018 MacBookAir on Jan 6, 2020
Processor: 1.6 GHz Intel Core i5
Memory: 8GB 2133 MHz LPDDR3
OS: macOS Mojave v10.14.6
Logical CPUs: 4
Physical CPUs: 2
$ make bench
Run all tests including benchmarks
$ make all
Just run benchmarks
$ make bench
Just run tests
$ make test
The Apache 2.0 License. See LICENSE for more details.
Copyright (c) 2020 Matrix Profile Foundation and contributors.
Please cite this work using our Journal of Open Source Software article
Van Benschoten et al., (2020). MPA: a novel cross-language API for time series analysis. Journal of Open Source Software, 5(49), 2179, https://doi.org/10.21105/joss.02179
[1] Chin-Chia Michael Yeh, Yan Zhu, Liudmila Ulanova, Nurjahan Begum, Yifei Ding, Hoang Anh Dau, Diego Furtado Silva, Abdullah Mueen, Eamonn Keogh (2016). Matrix Profile I: All Pairs Similarity Joins for Time Series: A Unifying View that Includes Motifs, Discords and Shapelets. IEEE ICDM 2016.
[2] Yan Zhu, Zachary Zimmerman, Nader Shakibay Senobari, Chin-Chia Michael Yeh, Gareth Funning, Abdullah Mueen, Philip Berisk and Eamonn Keogh (2016). Matrix Profile II: Exploiting a Novel Algorithm and GPUs to break the one Hundred Million Barrier for Time Series Motifs and Joins. IEEE ICDM 2016.
[3] Hoang Anh Dau and Eamonn Keogh (2017). Matrix Profile V: A Generic Technique to Incorporate Domain Knowledge into Motif Discovery. KDD 2017.
[4] Chin-Chia Michael Yeh, Nickolas Kavantzas, Eamonn Keogh (2017).Matrix Profile VI: Meaningful Multidimensional Motif Discovery. ICDM 2017.
[5] Shaghayegh Gharghabi, Yifei Ding, Chin-Chia Michael Yeh, Kaveh Kamgar, Liudmila Ulanova, Eamonn Keogh (2017). Matrix Profile VIII: Domain Agnostic Online Semantic Segmentation at Superhuman Performance Levels. ICDM 2017.