Datashader is a graphics pipeline system for creating meaningful representations of large amounts of data. It breaks the creation of images into 3 main steps:
-
Projection
Each record is projected into zero or more bins, based on a specified glyph.
-
Aggregation
Reductions are computed for each bin, compressing the potentially large dataset into a much smaller aggregate.
-
Transformation
These aggregates are then further processed to create an image.
Using this very general pipeline, many interesting data visualizations can be created in a performant and scalable way. Datashader contains tools for easily creating these pipelines in a composable manner, using only a few lines of code.
Datashader is available on most platforms using the conda
package manager,
from the bokeh
channel:
conda install -c bokeh datashader
Alternatively, you can manually install from the repository:
git clone https://github.com/bokeh/datashader.git
cd datashader
conda install -c bokeh --file requirements.txt
python setup.py install
Datashader is not currently provided on pip/PyPI, to avoid broken or low-performance installations that come from not keeping track of C/C++binary dependencies such as LLVM (required by Numba).
One way to easily install datashader
and related GIS and visualization tools is to install the conda environment from the examples
directory of a local datashader repository clone:
cd examples
conda env create
source activate ds
Note on Windows to replace source activate ds
with activate ds
.
There are lots of examples available in the examples
directory, most of
which are viewable as notebooks on Anaconda Cloud.
Additional resources are linked from the [datashader documentation] (http://datashader.readthedocs.org), including API documentation and papers and talks about the approach.