A brief intro to pydeck

pydeck is made for visualizing data points in 2D or 3D maps. Specifically, it handles

rendering large (>1M points) data sets, like LIDAR point clouds or GPS pings
large-scale updates to data points, like plotting points with motion
making beautiful maps

Under the hood, it's powered by the deck.gl JavaScript framework.

pydeck is strongest when used in tandem with Pandas but doesn't have to be.

Please note that these demo notebooks are best when executed cell-by-cell, so ideally clone this repo or run it from mybinder.org.



In [ ]:

    
import pydeck as pdk
print("Welcome to pydeck version", pdk.__version__)

There are three steps for most pydeck visualizations

We'll walk through pydeck using a visualization of vehicle accident data in the United Kingdom.

1. Choose your data

Here, we'll use the history of accident data throughout the United Kingdom. This data set presents the location of every latitude and longitude of car accidents in the UK in 2014 (source).



In [ ]:

    
import pandas as pd

UK_ACCIDENTS_DATA = 'https://raw.githubusercontent.com/visgl/deck.gl-data/master/examples/3d-heatmap/heatmap-data.csv'

pd.read_csv(UK_ACCIDENTS_DATA).head()

2. Configure the visualization: Choose your layer(s) and viewport

pydeck's Layer object takes two positional and many keyword arguments:

First, a string specifying the layer type, with our example below using 'HexagonLayer'
Next, a data URL–below you'll see the UK_ACCIDENTS_DATA that we set above, but we could alternately pass a data frame or list of dictionaries
Finally, keywords representing that layer's attributes–in our example, this would include elevation_scale, elevation_range, extruded, coverage. pickable=True also allows us to add a tooltip that appears on hover.

layer = pdk.Layer(
    'HexagonLayer',
    UK_ACCIDENTS_DATA,
    get_position='[lng,lat]',
    elevation_scale=50,
    pickable=True,
    auto_highlight=True,
    elevation_range=[0, 3000],
    extruded=True,                 
    coverage=1)

There is of course an entire catalog of layers which you're welcome to check out within the deck.gl documentation.

Configure your viewport

We also have to specifiy a ViewState object.

The ViewState object specifies a camera angle relative to the map data. If you don't want to manually specify it, the function pydeck.data_utils.compute_view can take your data and automatically zoom to it.

pydeck also provides some controls, most of which should be familiar from map applications throughout the web. By default, you can hold out and drag to rotate the map.



In [ ]:

    
layer = pdk.Layer(
    'HexagonLayer',
    UK_ACCIDENTS_DATA,
    get_position='[lng, lat]',
    auto_highlight=True,
    elevation_scale=50,
    pickable=True,
    elevation_range=[0, 3000],
    extruded=True,                 
    coverage=1)

# Set the viewport location
view_state = pdk.ViewState(
    longitude=-1.415,
    latitude=52.2323,
    zoom=6,
    min_zoom=5,
    max_zoom=15,
    pitch=40.5,
    bearing=-27.36)

# Combined all of it and render a viewport
r = pdk.Deck(layers=[layer], initial_view_state=view_state)
r.show()

Render an update to the visualization

Execute the cell below and look at the map in the cell above–you'll notice a seamless rendered update on the map



In [ ]:

    
layer.elevation_range = [0, 10000]

r.update()

Support updates over time

We can combine any Python function with our work here, of course. Execute the cell below to update our map above over time.



In [ ]:

    
import time
r.show()



In [ ]:

    
for i in range(0, 10000, 1000):
    layer.elevation_range = [0, i]
    r.update()
    time.sleep(0.1)

pydeck without Jupyter

If you prefer not to use Jupyter or you'd like to export your map to a separate file, you can also write out maps to HTML locally using the .to_html function.

(Note that if you're executing this example on mybinder.org, it won't render, since write access in the binder environment is restricted.)



In [ ]:

    
r.to_html('deck-iframe.html')