Fits Linear and Logistic Regression Models using MCMC.
Project description
# BayesianLinearRegression
This project was for me to gain a better understanding of the Metropolis-Hastings algorithm and work on my object-oriented programming skills. If you need to do any Bayesian modeling in Python, I recommend using PyMC3 (https://docs.pymc.io/).
#### – Project Status: [Completed]
## Project Intro/Objective
The purpose of this project was for me to work with the Metropolis-Hastings algorithm and get comfortable with object-oriented programming.
### Methods Used
Generalized Linear Models
Bayesian Statistics
Metropolis-Hastings Markov Chain Monte Carlo
Object-Oriented Programming
### Technologies
Python (NumPy, SciPy, tqdm)
## Project Description
I created a class called MetropolisHastingsLinearModel, which is the parent class of the GaussianModel, LaplacianModel, and LogisticModel classes. It has methods to calculate the log-prior (assuming normal priors for the distributions of the coefficients in the model), the log-posterior, fit the model/simulate the posteriors of the coefficients, burn the first x% the simulated distribution, set credible intervals, and predict for new data. For the GaussianModel and LaplacianModel classes, I only needed to add a log-likelihood method corresponding to those distributions. For the LogisticModel class, I added a method implementing the inverse-logit transformation for use in calculating the log-likelihood of the coefficients given the data and for predicting the probabilities of new observations. In addition to the method for predicting probabilities, I implemented a method for predicting the classes of new observations.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Hashes for BayesianLinearRegression-0.1.2.tar.gz
Algorithm | Hash digest | |
---|---|---|
SHA256 | 687084713cad48a0ccca4f644a48fb1250bfc8a22499c1c376df4b1f0841d3e4 |
|
MD5 | 83f22f1e854936f7630131074c7036e9 |
|
BLAKE2b-256 | 95b97e4eed62ddca5530dc2412cd2afb44dc4e1e65b6c8f4b2dec1e56ca63215 |