Workshop on introductory concepts in bayesian data analysis
The workshop comes in two slightly different versions:
Version presented at the IGGI Conference 2020
* Introduction
* Bayesian Gears
- From counts to probability
- Bayesian updating
- Likelihood, Parameters, Prior and Posterior
* Bayesian Machinery
- Parameters Estimation
- Grid Search, Quadratic Approximation, MCMC
* Bayesian Models
- PyMC3 Model Building
- Linear Regression
- Logistic Regression
- Graphical Models
Application-oriented version, more suitable for being delivered in industry settings
* Introduction
* Bayesian Approach to Inference
- Counts
- Updating Counts
- From Counts to Probability
- Likelihood, Parameters, Prior and Posterior
- Parameters Estimation
- Bayesian Models
* PyMC3
- Model Building
- Model Inspecting
- Model Fitting
- Model Evaluating and Comparing
- Model Predicting
* Applications
- PyMC3 vs scikit-learn
- Web Traffic Estimation
- Advertising Effect on Revenue
- Game Difficulty Estimation
- Model Comparison
- Statistical Rethinking (most of the content has been adapted from here)
- PyMC3
- Papers, Please (images and narrative framework for the University Workshop version)
- Download your local version of the workshop repository
- Install Anaconda
- If you run a Windows machine and do not have administrator rights:
- Open Anaconda Navigator
- Create a new environment with python=3.6 and call it workshop_env
- Open the Anaconda Powershell Prompt associated to the new environment
- Navigate to the workshop directory
- If you run a Windows machine and do have administrator rights:
- Open the Anaconda Powershell Prompt in the workshop directory
- Then from the Prompt:
# create anaconda environment
conda create -n workshop_env python=3.6
# activate the environment
conda activate workshop_env
- At this point install all the requirements with:
# install the requirements
conda install -c conda-forge --file requirements.txt
# open jupyter
jupyter notebook
- Navigate to and open the .ipynb file of interest
- Read the
Usage
section in the RISE documentation for navigating the notebook slides.