Running Lightning without a server

Setup



In [1]:

    
from lightning import Lightning

from numpy import random

Start local mode

Lightning was designed as a API-based visualization server, to which data is posted, and from which visualizations are returned. However, there are many use cases where operating without a server is desirable. For example, when doing data analysis locally, or when we're using notebooks like Jupyter.

For this use case, Lightning offers a "local" mode that doesn't require a server, or even internet access. This is a particularly easy way to get started with Lightning because it only requires a client installation! Once you've installed the Python client with pip, all you need to do is set local mode to true.



In [2]:

    
lgn = Lightning(ipython=True, local=True)









    




Lightning initialized






    











    



Running local mode, some functionality limited.

Generate a plot

Then generate a plot. It'll automatically embed in the notebook (because we set ipython=True). Local plots are interactive just like plots rendered using the server! Try zooming and panning.



In [3]:

    
series = random.randn(5, 50)

lgn.line(series)









    Out[3]:

Performance can often be a little better for local plots with large data sets, because there is no data transfer.



In [4]:

    
x = random.randn(1000)
y = random.randn(1000)
v = random.randn(1000)

lgn.scatter(x, y, alpha=0.5, values=v, colormap='Reds')









    Out[4]:

And most plot types are available. Here's a force visualization.



In [5]:

    
mat = random.rand(100,100)
mat[mat<0.97] = 0

lgn.force(mat)









    Out[5]:

And here's a map!



In [6]:

    
states = ["NA", "AK", "AL", "AR", "AZ", "CA", "CO","CT",
          "DC","DE","FL","GA","HI","IA","ID","IL","IN",
          "KS","KY","LA","MA","MD","ME","MI","MN","MO",
          "MS","MT","NC","ND","NE","NH","NJ","NM","NV",
          "NY","OH","OK","OR","PA","RI","SC","SD","TN",
          "TX","UT","VA","VI","VT","WA","WI","WV","WY"]
values = random.randn(len(states))

lgn.map(states, values, colormap='Greens')









    Out[6]:

Saving to html

You can also save a visualization to html, which is useful if you are using local mode without a notebook. First create the visualization.



In [7]:

    
viz = lgn.scatter(random.randn(10), random.randn(10))

Then save using viz.save_html('filename')

Limitations

Some visualizations are not available in local mode, for example, plots that use images (though we are working on expanding coverage).



In [8]:

    
lgn.image(random.randn(25,25))









    



Plots of type 'image' not yet supported in local mode

And although full interactivity is supported, there is currently no way to extract user selections from a visualization. For that, take a look at the various ways of running Lightning with a server!