\\n\"+\n",
" \"
\\n\"+\n",
" \"BokehJS does not appear to have successfully loaded. If loading BokehJS from CDN, this \\n\"+\n",
" \"may be due to a slow or bad network connection. Possible fixes:\\n\"+\n",
" \"
\\n\"+\n",
" \"
\\n\"+\n",
" \"- re-rerun `output_notebook()` to attempt to load from CDN again, or
\\n\"+\n",
" \"- use INLINE resources instead, as so:
\\n\"+\n",
" \"
\\n\"+\n",
" \"
\\n\"+\n",
" \"from bokeh.resources import INLINE\\n\"+\n",
" \"output_notebook(resources=INLINE)\\n\"+\n",
" \"\\n\"+\n",
" \"
![](\"assets/images/bokeh-transparent.png\"){kind=spacer}
\n",
" \n",
" ![](\"assets/datashader_examples.png\")
"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## When *not* to use datashader\n",
"\n",
"* Plotting less than 1e5 or 1e6 data points\n",
"* When *every* datapoint matters; standard Bokeh will render all of them\n",
"* For full interactivity (hover tools) with *every* datapoint\n",
"\n",
"## When *to* use datashader\n",
"\n",
"* Actual *big data*; when Bokeh/Matplotlib have trouble\n",
"* When the *distribution* matters more than individual points\n",
"* When you find yourself sampling or binning to better understand the *distribution*\n",
"\n",
"\n",
"# How does datashader work?\n",
"\n",
"![](\"https://raw.githubusercontent.com/bokeh/datashader/master/docs/images/pipeline2.png\")
\n",
"\n",
"* Tools like Bokeh map Data directly into an HTML/JavaScript Plot\n",
"* datashader renders Data into a screen-sized Aggregate array, from which an Image can be constructed then embedded into a Bokeh Plot \n",
"* Only the fixed-sized Image needs to be sent to the browser, allowing millions or billions of datapoints to be used\n",
"* Every step automatically adjusts to the data, but can be customized\n",
"\n",
"# Visualizations supported by datashader\n",
"\n",
"Datashader currently supports:\n",
"\n",
"* Scatterplots/heatmaps\n",
"* Time series\n",
"* Connected points (trajectories)\n",
"* Rasters\n",
"\n",
"In each case, the output is easily embedded into Bokeh plots, with interactive resampling on pan and zoom, in notebooks or apps. Legends/hover information can be generated from the aggregate arrays, helping provide interactivity.\n",
"\n",
"# Faithfully visualizing big data\n",
"\n",
"Once data is large enough that individual points are not easily discerned, it is crucial that the visualization be constructed in a principled way, faithfully revealing the underlying distribution for your visual system to process. For instance, all of these plots show the same data -- is any of them the real distribution?\n",
"\n",
"![](\"assets/5_gaussians_3_ways.png\")
"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Let's find out! The data in the above images was created by summing five normal (Gaussian) distributions as follows:"
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"collapsed": false
},
"outputs": [
{
"data": {
"text/html": [
"![](\"assets/5_gaussians_labeled.png\")
\n",
"\n",
"Let's look at each of the stages in the datashader pipeline in turn, to see how images like this are constructed and how they can be controlled and embedded into Bokeh plots.\n",
"\n",
"# Projection and Aggregation\n",
"\n",
"The first stages of the datashader pipeline are to choose:\n",
"\n",
"* which variables you want to plot on the x and y axes,\n",
"* what size array do you want to aggregate the values into,\n",
"* what range of data values should that array cover, and\n",
"* what \"reduction\" function you want to use for aggregating:"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"collapsed": false
},
"outputs": [
{
"data": {
"text/plain": [
"