You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
BAMT 2.0.0 - new features, refactoring, architecture refreshment #108
Current BAMT architecture has a number of disadvantages, some clunky code and other limitations. Thus, it was decided to make a full refactoring. This refreshment will not only include new refactored code and API but also new features (like vectorized sampling and other operations, new algorithms for structure learning, score-functions etc.)
For now, here is a checklist of modules that should be implemented in 2.0.0 architecture:
The development of BAMT 2.0.0 is held in 2.0.0 branch of the repository. If you, the reader of the issue, have decided to implement some module or submodule, please reply to this message, create a separate issue and add it to milestone and project.
The goal of these changes is also to make a sklearn-like interface, use modern python development practices like Pydantic, so the usual pipeline looks like that:
# read datadata=pd.read_csv("data.csv")
# define optimizers and score functionsdag_score_function=DAGScoreFunction(**parameters)
dag_optimizer=DAG_optimizer(**parameters)
# get a structure, maybe in networkx format?G=dag_optimizer.optimize(data, **parameters)
# define parameters estimator and BNparameters_estimator=ParametersEstimator(**parameters)
bn=ContinuousBayesianNetwork(**parameters)
# fit the bnbn.fit(data, ParametersEstimator, **parameters)
bn.sample(1000)
bn.predict(data.drop[["col1", "col2"]])
Hello! I've analyzed the current state of the 2.0.0 refactoring and I'm ready to contribute.
Current Status Analysis
I've examined the 2.0.0 branch and found that most of the architecture skeleton is in place:
✅ Implemented (skeleton/abstract classes):
Core module (Graph, DAG, Nodes, Node Models)
DAG optimizers (constraint-based, score-based, hybrid)
Score functions (K2, Mutual Information - both are stubs)
Parameter estimators (MLE)
Models (all BN types)
❌ Missing from checklist:
BIC/AIC score functions (mentioned but not implemented)
Actual implementations for most skeleton classes
Question
Since this is a meta-issue for the overall 2.0.0 refactoring, could you please clarify which specific module(s) would be most valuable for me to implement? Some options:
BIC/AIC Score Functions - These are explicitly mentioned in the checklist as missing
Complete implementation of existing stubs - Many classes like K2Score, MutualInformationScore are just empty implementations
Other specific module - If there's a particular area that's a priority
I'm ready to create a separate sub-issue and add it to the milestone once we determine the focus area.
Hello! I've analyzed the current state of the 2.0.0 refactoring and I'm ready to contribute.
Current Status Analysis
I've examined the 2.0.0 branch and found that most of the architecture skeleton is in place:
✅ Implemented (skeleton/abstract classes):
Core module (Graph, DAG, Nodes, Node Models)
DAG optimizers (constraint-based, score-based, hybrid)
Score functions (K2, Mutual Information - both are stubs)
Parameter estimators (MLE)
Models (all BN types)
❌ Missing from checklist:
BIC/AIC score functions (mentioned but not implemented)
Actual implementations for most skeleton classes
Question
Since this is a meta-issue for the overall 2.0.0 refactoring, could you please clarify which specific module(s) would be most valuable for me to implement? Some options:
BIC/AIC Score Functions - These are explicitly mentioned in the checklist as missing
Complete implementation of existing stubs - Many classes like K2Score, MutualInformationScore are just empty implementations
Other specific module - If there's a particular area that's a priority
I'm ready to create a separate sub-issue and add it to the milestone once we determine the focus area.
@konard You have to implement all the features present in the current main branch in the new architecture. There is no need for sub-issues, or anything like that. Just add a complete refactoring to your new pull request.
Activity
jrzkaminski commentedon Jul 5, 2024
distribution module has been implemented
Initial commit with task details for issue aimclub#108
konard commentedon Nov 13, 2025
Hello! I've analyzed the current state of the 2.0.0 refactoring and I'm ready to contribute.
Current Status Analysis
I've examined the
2.0.0branch and found that most of the architecture skeleton is in place:✅ Implemented (skeleton/abstract classes):
❌ Missing from checklist:
Question
Since this is a meta-issue for the overall 2.0.0 refactoring, could you please clarify which specific module(s) would be most valuable for me to implement? Some options:
I'm ready to create a separate sub-issue and add it to the milestone once we determine the focus area.
Looking forward to your guidance!
Related: PR #141
Revert "Initial commit with task details for issue aimclub#108"
jrzkaminski commentedon Nov 13, 2025
@konard You have to implement all the features present in the current main branch in the new architecture. There is no need for sub-issues, or anything like that. Just add a complete refactoring to your new pull request.
feat: Implement score functions (K2, MI, BIC, AIC) for new 2.0.0 arch…
feat: Implement Hill Climbing optimizer for structure learning
feat: Implement core 2.0.0 components - node models, conditional node…
feat: Add score functions, continuous/hybrid BN implementations, and …
feat: Add utilities and PC optimizer for BAMT 2.0.0 architecture
feat: Add MixtureGaussianDistribution and MathUtils for BAMT 2.0.0
feat: Add extended ML models repository and example for BAMT 2.0.0