flot-downsample: Downsample plugin for Flot

Author: Sveinn Steinarsson
Special thanks: Borgar Þorsteinsson and DataMarket

This plugin for Flot charts downsamples
data before rendering the chart. The purpose is to try retain the
visual characteristics of the original line using considerably fewer data points.
The algorithm (called Largest-Triangle-Three-Buckets or LTTB) used in this plugin is described in a
Master’s thesis (see hdl.handle.net/1946/15343) in Computer Science by Sveinn Steinarsson
at the University of Iceland.
The topic of the thesis is how to downsample time series for visual representation
and was initially suggested by DataMarket.
JavaScript optimization was done with the help of Borgar Þorsteinsson.

Status

The Flot library is not commonly used nowadays, but the LTTB (Largest-Triangle-Three-Buckets) algorithm remains popular and is well-established for downsampling chart data.

For using LTTB in other libraries or frameworks, you can refer to the LTTB adapted for other programming languages or frameworks section.

Demo

Demo can viewed at base.is/flot/.

Additional demo concerning chart resizing is also available at base.is/flot/resize/

Usage

series: {
  downsample: {
    threshold: 1000 // 0 disables downsampling for this series.
  }
}

Known limitations

Does not support gaps (null values) in the data array.
X-values must be in a strictly increasing order.

The Billion Data Point Challenge by Benjamin Raskin and Nikunj Aggarwal at Uber Technologies Inc.
@hayley.morrison/sampling-time-series-data-sets-fc16caefff1b">An SQL implementation of the ‘Largest Triangle Three Bucket’ algorithm by Peter Coppens
Visualize Big Data on Mobile by dduraz
The Challenge of Metrics Visualization by Mark Troyer
Downsampling data – Not a trivial task by DataMarket (The company was acquired by Qlik. Linking to archived version since the original blog is no longer accessible)
Sampling large datasets in d3fc by William Ferguson
Downsampling algorithms by Adrian S. Tam
Plot your data faster without losing its shape with Elixir and ExLTTB by Riccardo Binetti
Implementation of Downsampling Algorithm in MSChart Extension by Code Artist
MinMaxLTTB: Leveraging MinMax-Preselection to Scale LTTB by Jeroen & Jonas Van Der Donckt

The algorithm (LTTB) adapted for other programming languages or frameworks

C# by Daniel Judge
C# by Adrian Seeley
MSChartExtension in C# by Code Artist
Elixir by Riccardo Binetti
Go by Damian Gryski
node.js by Matej Drolc
Java by drcrane
Java8 by Guillermo Gutierrez Almazor
Scala by Graham Tackley
R by Jack Viljoen
Javascript (without the flot plugin) by Josh Carr
Highcarts Plugin by Sveinn Steinarsson
Chart.js Plugin by Sean
PHP by dbojdo
Python by Olivier Devoisin
Python + Numpy by Jack Viljoen
Python using C by dgoeries
Python using Rust by Jeroen Van Der Donckt
Perl by troxel
C++ by parkertomatoes
Ruby by Julian Lübke
Swift by Guillaume Beal
Rust by Jerome Froelich
Clojure by Kimmo Koskinen
Dart by bnap00
PL/PGSQL by Rio Fujita
PostgreSQL Extension by Brian Rowe

Note: I have not tested all these versions and some might have different or additional features.

(Please let me know if you make your own port of the LTTB algorithm so I can list it here.)

License

flot-downsample is released under the terms of the MIT License.