iventure serves as a jupyter-based front-end for BayesDB.

For access to iventure, please refer to the release webpage of the Probabilistic Computing Stack.

Please refer to the following tutorial notebooks for illustrative probabilistic data analysis tasks on real-world datasets:

• Exploratory analysis on Gapminder, a dataset of global macroeconomic indicators of education, poverty, environment and health.

• Predictive analysis on a table of Earth satellites from the Union of Concerned Scientists.

The magics.py contains cell magics which allow for interactive probabilistic programming in a Jupyter notebook.

In the first cell of a Jupyter notebook, load the extension.

%load_ext iventure.magics

%bayesdb foo.bdb

%mml CREATE POPULATION ...

or, for multi-line schemas.

%%mml
DROP POPULATION xyz;
CREATE POPULATION xyz for t (...);

%bql ESTIMATE MUTUAL INFORMATION OF x WITH y WITHIN xyz;

or, for multi-line queries

%%bql
CREATE TEMP TABLE depprobs AS
    ESTIMATE DEPENDENCE PROBABILITY FROM PARIWSE VARIABLES OF xyz;
.plot SELECT * FROM depprobs

%bql .nullify satellites_t NaN
%bql .population satellites_p

%bql .scatter SELECT apogee_km, perigee_km FROM satellites_t LIMIT 100;