Automatically train a merged Bayesian Network from multiple data sources
Using pip: open autopgm
folder, and run
$ pip3 install .
In python3
, run
from autopgm.estimator import MultipleBayesianEstimator
model = MultipleBayesianEstimator([csv_file_name_1, csv_file_name_2, ...]).get_model()
Note: all files need to be .csv files with discrete variables of integer values.
model
is a BayesianModel
as in pgmpy
.
You can perform VariableElimination
and then query
as in pgmpy
:
from pgmpy.inference import VariableElimination
inference = VariableElimination(model)
q = inference.query(['var1'], evidence={'var2': 0, 'var3': 1})['var1']