Ankur Ankan

Bug in EM when latent variables are present

3

### Subject of the issue EM algorithm seems to be assigning equal probability values to all the states of latent variables. ### Your environment * pgmpy version: 0.1.18 * Python...

Batch query for inference methods

1

### Your checklist for this pull request Please review the [guidelines for contributing](CONTRIBUTING.md) to this repository. - [ ] Make sure you are requesting to **pull a topic/feature/bugfix branch** (right...

Out of memory operations on huge numpy arrays

14

### Subject of the issue The size of numpy arrays can get quite huge in `DiscreteFactor` resulting in `MemoryError`. Therefore, we need a way to store these arrays out of...

Parameter and Structure learning for continuous data

11

Should let the user specify what kind of distribution the data should be assumed to be coming from. So for Bayesian Learning we can have the API like: ```python from...

### Subject of the issue The closure method doesn't seem to give correct values in some cases. ### Your environment * pgmpy version: dev * Python version: 3.6 * Operating...

Add type hinting in the codebase

2

From python 3.4, type hinting feature has been added to specify the expected type of arguments in functions/methods/classes and their returns types. Modify our codebase to have type hintings as...

Genetic Algorithm: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=537345 Ant Colony Optimization: https://www.sciencedirect.com/science/article/pii/S0888613X02000919 Review: https://www.cambridge.org/core/services/aop-cambridge-core/content/view/146D29D291B7393EF62BAFE1D47D4426/S0269888910000251a.pdf/learning-bayesian-networks-approaches-and-issues.pdf

The current implementation of causal inference functionality is limited to DAGs but more generalized representations have been used in literature. - [ ] Ancestoral Graphs (directed mixed graphs) - [...

Robust tests for readwrite module

1

The tests for `pgmpy.readwrite` modules are quite basic. More tests need to be added particularly to check the following: - [ ] Different order of inputs results in the same...

### Subject of the issue Some of the methods in causal inference right now iterate over all possible paths or all possible variable combinations for finding d-separation variables. This is...