https://github.com/ying531/MCMC-SymReg

Symbolic Regression using MCMC sampling
https://github.com/ying531/MCMC-SymReg

Last synced: 4 months ago
JSON representation

Symbolic Regression using MCMC sampling

• Host: GitHub
• URL: https://github.com/ying531/MCMC-SymReg
• Owner: ying531
• Created: 2019-05-30T03:11:41.000Z (about 5 years ago)
• Default Branch: master
• Last Pushed: 2022-01-22T08:19:16.000Z (over 2 years ago)
• Last Synced: 2024-01-16T10:44:03.075Z (5 months ago)
• Language: Python
• Homepage:
• Size: 1.94 MB
• Stars: 21
• Watchers: 1
• Forks: 7
• Open Issues: 2
• Metadata Files:
  • Readme: README.md

Lists

• awesome-stars - ying531/MCMC-SymReg - Symbolic Regression using MCMC sampling (Python)

README

        
# MCMC-SymReg

Bayesian symbolic regression using mcmc sampling. 

Paper here: https://arxiv.org/abs/1910.08892

## API and Usage

### Codes location

`codes/BSR.py`: API interface of BSR class

`codes/bsr_class.py`: definition of BSR class

`codes/simulations.py`: part of simulation settings in the paper

`codes/funcs.py`: basic sampling functions

### Usage Example

```python

K = 3 # number of trees

MM = 50 # number of iterations

# set hyperparameters alternatively

hyper_params = [{'treeNum': 3, 'itrNum':50, 'alpha1':0.4, 'alpha2':0.4, 'beta':-1}]

# initialize BSR object

my_bsr = BSR(K,MM)

# train (need to fill in parameters)

# train_X is dataframe with each row a datapoint

# train_y is series with default index

my_bsr.fit(train_X,train_y)

# fit new values

# new_X is dataframe of new data

fitted_y = my_bsr.predict(new_X)

# display fitted trees

my_bsr.model()

# complexity, including complexity of each tree & total

complexity = my_bsr.complexity()

```

>>>>>>> 

## Pdf files

`bsr_paper.pdf`: paper for Bayesian Symbolic Regression

`Symbolic_Regression_Tree_MCMC.pdf`: note for proposed algorithm