CODEWITHSUNDEEP
machine-learningdatasetbayesian-networks
A. May
A. May

Reputation: 69

Dataset for Bayesian Network Structure Learning

After tons of researches, I didn't find a repository with the necessary material to test a algorithm able to learn the structure of a Bayesian Network. What I need are only 2 things:

My algorithm should be able to learn the structure from the dataset and then I could check how far from the right BN it is. Do you have any links? I've already found some dataset without the original BN and viceversa but I need both of them for my university project.

Thanks in advance

PS: if you are interested, I use Python for my project.

Upvotes: 1

Views: 1339

Answers (1)

erdogant
erdogant

Reputation: 1694

Try the bnlearn library. It contains structure learning, parameter learning, inference and various example datasets such as sprinkler, asia, alarm, and many more.

  • Various examples can be found here.
  • Blog about detecting causal relationships can be found here.

Example for structure learning and making inferences:

# Load library
import bnlearn as bn

# Load Asia DAG
DAG = bn.import_DAG('asia')

# plot ground truth    
G = bn.plot(DAG)

# Sampling
df = bn.sampling(DAG, n=10000)

# Structure learning
model_sl = bn.structure_learning.fit(df, methodtype='hc', scoretype='bic')

# Plot based on structure learning of sampled data
bn.plot(model_sl, pos=G['pos'], interactive=True)

# Compare networks and make plot
# bn.compare_networks(model, model_sl, pos=G['pos'])

Upvotes: 1

Related Questions