Speaker: Thomas Wiecki
Title: The Bayesian Workflow: Building a COVID-19 Model (Part 1)
Event description:
In this tutorial we will build a COVID-19 model from scratch.
Discourse Discussion
https://discourse.pymc.io/t/the-bayesian-workflow-building-a-covid-19-model-by-thomas-wiecki/6017
## Timestamps
00:00 Speaker’s introduction
00:58 Key strengths of bayesian statistics
03:29 Agenda - what will you learn today
04:00 Dataset
05:24 Bayesian workflow
06:52 Plot the data for Germany cases
08:11 Instantiate model and set parameters for exponential regression
12:26 Run prior predictive check
14:24 Fit model
15:00 Mass matrix traceback and sampling issues
18:35 Proposing a better model - update parameters and likelihood distribution
20:54 Fit the new model and assess convergence
22:46 Run posterior predictive check
24:35 Prediction and Forecasting with pm.Data
25:21 Update intercept and slope parameters
26:14 Update model data using pm.Data container
27:24 Plot results and discuss model quality
28:35 Improve model by fitting a Logistic regression
31:49 Compare models
33:39 Fit the logistic regression to US data
34:20 Model main limitations and topic for next video
Speaker bio:
- Thomas is the founder of PyMC Labs, a Bayesian consulting firm.
- PyMC author
- PhD on computational cognitive neuroscience from Brown University
- Former VP of data science and head of research at Quantopian Inc: building a team of data scientists to build a hedge fund from a pool of 300k crowd researchers.
- Recognized public speaker: keynotes at the Open Data Science Conference (ODSC) & TACC as well as talks at Strata Hadoop & AI, Newsweek AI conference, and various PyData conferences around the world
- Runs a data science podcast, blogs about Bayesian statistics, and is an avid Twitter personality.
- Data scientist to follow of the year 2015, ODSC Open Source award 2018.
Speaker info:
Website: https://twiecki.io
GitHub: https://github.com/twiecki
Twitter: https://twitter.com/twiecki
LinkedIn: https://www.linkedin.com/in/thomas-wiecki-46339244/
Part of PyMCon2020.
More details at http://www.pymcon.com
#bayesian #statistics #covid19