My GSoC 2016 work can be found here:

• https://github.com/pgmpy/pgmpy/commits?author=chrisittner

In addition, these two Pull Requests are not yet merged:

• https://github.com/pgmpy/pgmpy/pull/720
• https://github.com/pgmpy/pgmpy/pull/721

I wrote an introduction to pgmpy’s new structure learning ...

GSoC 2016 is coming to an end I’ve just sent the last PR necessary to complete the scope of my proposal. It has been an exciting project, and I do feel that I learned a lot. I was able to implement a number of basic BN structure estimation algorithms ...

The past while I have been working on basic constraint-based BN learning. This required a method to perform conditional independence tests on the data set. Surprisingly, such tests for conditional independence are not part of scipy.stats or other statistics libraries.

To test if X _|_ Y | Zs, one ...

pgmpy now has a basic hill climb BN structure estimator.

Usage:

import pandas as pd
import numpy as np
from pgmpy.estimators import HillClimbSearch, BicScore

# create data sample with 9 random variables:
data = pd.DataFrame(np.random.randint(0, 5, size=(5000, 9)), columns=list('ABCDEFGHI'))
# add 10th dependent variable ...

After some doubts about the BN structure learning performance on a few test cases, I cross checked my structure scoring functions against the implementation from the bnlearn R package. After minor changes the implemented scores now yield the exact same results:

pgmpy:

~$ python
Python 3.5.2 (default, Jun 28 ...

This is a (possibly already outdated) summary of structure learning capabilities of existing Python libraries for general Bayesian networks.

libpgm.pgmlearner

• Discrete MLE Parameter estimation
• Discrete constraint-based Structure estimation
• Linear Gaussian MLE Parameter estimation
• Linear Gaussian constraint-based Structure estimation

Version 1.1, released 2012, Python 2

bnfinder (also here)

• Discrete ...

I’ll soon finish basic score-based structure estimation for BayesianModels. Below is the current state of my PR, with two examples.

Changes in pgmpy/estimators/

I rearranged the estimator classes to inherit from each other like this:

.                                    MaximumLikelihoodEstimator
                                   /
                ParameterEstimator -- BayesianEstimator
              /
BaseEstimator                        ExhaustiveSearch
            | \                    /
            |   StructureEstimator -- HillClimbSearch
            |                      \
            |                        ConstraintBasedEstimator
            |
            |
            |                BayesianScore
            |              /
            StructureScore -- BicScore

BaseEstimator ...

With a bit of delay, I am now working on a basic PR for score-based structure estimation for Bayesian Networks. It comes with two ingredients:

• The StructureScore-Class and its subclasses BayesianScore and BICScore. They are initialized with a data set and provide a score-method to compute how well ...

Now that ML Parameter Estimation works well, I’ve turned to Bayesian Parameter Estimation (all for discrete variables).

The Bayesian approach is, in practice, very similar to the ML case. Both involves counting how often each state of the variable obtains in the data, conditional of the parents state. I ...

At the moment pgmpy supports Maximum Likelihood Estimation (MLE) to estimate the conditional probability tables (CPTs) for the variables of a Bayesian Network, given some data set. In my first PR, I’ll refactor the current MLE parameter estimation code to make it a bit nicer to use. This includes ...

My proposal for Google Summer of Code 2016 has been accepted :). This means that I will spend part of my summer working on the pgmpy library. I will implement some techniques for Bayesian Network structure learning. You can have a look an my proposal here.

As a first step, I ...