Exoplanets example


This applet is intended as a sandbox for testing PFA functions on a real dataset. The data are 1783 extrasolar planets, downloaded in April 2014 from exoplanet.eu.

The dataset has a two-level taxonomy: it is a collection of stars and each star has one or more planets. All of the numerical data for stars and planets are nullable (missing data is common in planet-hunting). A few fields are strings (names and spectral type), one is an enumerated category (detection type), and one is an array of strings (molecules discovered in the planet’s atmosphere).

Rather than dump thousands of JSON-formatted outputs in your browser, the output is drawn as a scatter plot using d3. Don’t read too much into the distribution; extrasolar planets have a strong detection bias.

The dataset is hosted on Google App Engine (the same site that executes the PFA) as a binary Avro file. If Google App Engine has to start a new instance, the calculation will take several seconds. Otherwise, most runs will be less than a second.


The default executable is a fold-type scoring engine that only produces output for planets when planet.radius, planet.mass, star.age, star.temp, and star.dist are all non-null. This is not required for the example to work; any action will do, as long as the input and output types are unchanged. (The input must match the exoplanets dataset and the output must be correctly interpreted by your browser.)