Vegas_spark_charting_library

vegas-viz.org

passion project from two guys at Netflix

charting with scala

want to be able to aggregate, filter, facet, color

it can do all of that in one step --> javascript --> rendering 

looks like a very concise version of D3 but including the filtering, faceting, and aggregation

plus, because it's D3 you get interactivity (or it's coming soon, anyway)

anatomy of a plot: channels
x/y channel, shape channel, color channel, size channel 

vegas supports all the basic chart types: 
bar, scatter, area, streams
trellising and layers (bar chart + area for error + triangles)

runs in Zeppelin notebook and consoles (scala REPL will pop up a window and display the plot)

should work on Databricks, Jupyter

1st class support for spark dataframes - it knows if you give it a column name, what that means

visual statistics:
binning, sorting scaling, transforms, time series, aggregation, filtering, support for missing data 
(what does that mean, I don't have to remove NaNs like I do for matplotlib?)

1. write the scala
2. chart is rendered via javascript embedded in an HTML iFrame

vegas (scala emits type-checked vega-lite json) --> vega-lite (converts internally to vega json spec) 
--> vega (translates to d3js json) --> d3js

someone asked if the aggregations are actually done in spark - the answer is no, it's done in the browser, 
they expect you to have pre-filtered the data

q: how much data can it handle? ~200k rows

q: how easy is it to change font sizes, orientation of labels, location of legend? - everything Vega-lite supports, this supports

coming soon: interactive selections with a list or a paintbrush (Halelujah!) - including linked (demo looks good so far!) 

*he mentioned a google widget for this?

they said this is very similar to Altair

someone asked about if you resize/zoom interactively, can it save that (answer is no, but they thought that was an interesting suggestion)

someone asked how hard it is to get it going - only need 1 jar to use (I think that's what he said)

someone else asked re: live streaming - answer is not supported yet