Numpy implementation of Steinarsson’s Largest-Triangle-Three-Buckets algorithm for downsampling time series–like data

Based on the original JavaScript code:

https://github.com/sveinn-steinarsson/flot-downsample

and:

Sveinn Steinarsson. 2013. Downsampling Time Series for Visual Representation. MSc thesis. University of Iceland.

A test data set is provided in tests/timeseries.csv. It was downloaded from http://flot.base.is/ and converted from JSON to CSV.

import numpy as np
import lttb
data = np.array([range(100), np.random.random(100)]).T
small_data = lttb.downsample(data, n_out=20)
assert small_data.shape == (20, 2)

For example, here is the data set provided in tests downsampled to 100 points:

Download the source code and install the lttb package into your (virtual) environment:

git clone https://github.com/javiljoen/lttb.py.git
pip install ./lttb.py

• Python 3
• Numpy

Licence: MIT