A course by Peadar Coyle, a core PyMC3 contributor and founder of Aflorithmic Labs
Overview
Bayesian methods are powerful tools for data science applications, complimenting traditional statistical and machine learning methods. Importantly, Bayesian models generate predictions and inferences that fully account for uncertainty. The main tool for conducting Bayesian analysis is Markov chain Monte Carlo (MCMC), a computationally-intensive numerical approach that allows a wide variety of models to be estimated. MCMC algorithms are available in several Python libraries, including PyMC3. I will teach users a practical, effective workflow for applying Bayesian statistics using MCMC via PyMC3 using real-world examples.
This course is intended for analysts, data scientists and machine learning practitioners. Anyone looking for effective ways of making predictions and obtaining inference from datasets should find it useful. The material will assume an intermediate level of Python familiarity. Ideally, attendees should be familiar with Numpy and Jupyter. There is no expectation of students having a statistical background.
The key learnings from the course will be
- What PyMC3 is for
- What MCMC is and why should I care
- How to know enough theano to not be scared by it
- How to diagnose things like model convergence and figure out if your model is good or not
- An introduction to Multi-level models or Bayesian Machine Learning secret sauce
- Contains 4 real world case studies including sports analytics, AB testing, policy analysis and predicting process anomalies
- Exclusive walkthroughs of other Bayesian Machine Learning tools like Arviz, Rainier and Pyro
Having completed the course, students should be able to build basic Bayesian statistical models using their own data, validate those models, and interpret their output
This course is intended for analysts, data scientists and machine learning practitioners. Anyone looking for effective ways of making predictions and obtaining inference from datasets should find it useful. The material will assume an intermediate level of Python familiarity. Ideally, attendees should be familiar with Numpy and Jupyter. There is no expectation of students having a statistical background.
The key learnings from the course will be
- What PyMC3 is for
- What MCMC is and why should I care
- How to know enough theano to not be scared by it
- How to diagnose things like model convergence and figure out if your model is good or not
- An introduction to Multi-level models or Bayesian Machine Learning secret sauce
- Contains 4 real world case studies including sports analytics, AB testing, policy analysis and predicting process anomalies
- Exclusive walkthroughs of other Bayesian Machine Learning tools like Arviz, Rainier and Pyro
Having completed the course, students should be able to build basic Bayesian statistical models using their own data, validate those models, and interpret their output
Build robust models and interpret them
Bayesian Analysis provides robust ways of interpreting parameters. Such as which rugby team is the best at attacking!
What's included in the course?
- Introductions to Bayesian Statistics, PyMC3, Theano and MCMC. Including applications to Pyro, Rainier and ArviZ so you won't be constrained by PyMC3.
- Over 5 hours of professionally edited videos and quizzes to help you learn
- Descriptive Overviews of Core Models and the Value of Probabilistic Programming
- Walkthrough Videos That Show You Exactly How to Build and Debug these models.
- Documents and Notebooks Designed to Help you upskill and understand the technical underpinnings and how Probabilistic Programming relates to Deep Learning. These are based on hours of lectures and workshops internally at top startups and at major Data Science conferences.
- Quizzes - Quizzes so you can make sure you understand the material
- Guidance and Support From a Large (and Growing!) Community of Like-minded Data Scientists
- Lifetime Access to Our Private Slack Community of Over 50 Ambitious Data Scientists and some core contributors of PyMC3 (£200/year value)
- Over 20 screencasts - Screencasts taking you through both the theory and the implementation of Probabilistic Programming.
- Real world examples - This course is about practicality. It includes real world case studies including the safety of self driving cars, how to use Bayesian Statistics to optimise a supply chain, Bayesian AB testing and applications to Sports Analytics.
Learn how to convert your model into $
I'll teach you how to evaluate and trust your model. And more importantly convert that into money for your company!
What course students are saying
Unlike academia or blogs which focus solely on theory or application, Peadar combines both in those course to set a solid foundation for his students. With the knowledge from this course students will be empowered in Bayesian methods, whether they want to read papers, or start applying the methods in PyMC3 themselves
Ravin Kumar - Engineer and PyMC3 contributor
The probabilistic programming primer is an incredible course that offers a fast track to an incredibly exciting field. Peadar clearly communicates the content and combines this with practical examples which makes it very accessible for his students to get started with probabilistic programming.
Peter Verheijen - Entrepreneur
Other talks - Conferences
I've been giving talks about Bayesian Statistics to audiences around the world for a few years now.
This is one on Probabilisitic Programming that's worth a look it was to a packed audience in London.
This is one on Probabilisitic Programming that's worth a look it was to a packed audience in London.
What's included?
33 videos
17 files
2 text files
Contents
Hands on introduction to PyMC3
What is PyMC3 and how do I get started?
What is MCMC?
Introduction to Theano
Hierarchical Models (Multilevel models)
Model building with PyMC3
Other Probabilistic Programming Languages and Tools
Real world examples of Probabilistic Programming
Bayesian Machine Learning
FAQ
Will there be course support?
There is a Slack community and you get to join an exclusive network consisting of other Bayesian Statisticians.
I'll also do some Q and A sessions during the next few months so people can ask me questions and I'll answer them. The video will be recorded and added to the course.
I'll also do some Q and A sessions during the next few months so people can ask me questions and I'll answer them. The video will be recorded and added to the course.
Is there a money back guarantee?
We'll offer 60 day money back guarantee under the provision that you show some evidence of attempting some of the homework exercises.
Just email me and we'll sort out a refund.
Just email me and we'll sort out a refund.
What applications in the real world are there for Probabilistic Programming?
In my own career I've applied Bayesian Statistics (Probabilistic Programming) in Marketing, Financial Services, Sports Analytics and Energy markets.
Some use cases I can imagine my audience being interested in
Some use cases I can imagine my audience being interested in
- Modelling risk of financial or insurance products
- Anomaly detection in time series data - this could be for fraud, identity modelling, overloads in systems, manufacturing, causal models in marketing.
- Modelling the safety of self driving cars (which has applications in many policy use cases)
- Sports Analytics
- Modelling the inherent uncertainty in any measurement process - for example in a two sided recruitment market you might have scores assigned by job seekers to companies and by companies to job seekers
In summary - anywhere that uncertainty matters to you, or you have domain specific expertise that can be part of the modelling process could be amenable to Bayesian Statistics.
We know for a fact that PyMC3 and Stan are used at world leading companies such as Google, Facebook, Hotels.com, Quantopian, Channel 4, Monetate, Freebird, Harvard and Uber.
But Is This Course Worth It?
Years of experience and best practices packed into over 4 hours of content.
Instead of spending hundreds of hours furiously Googling around on your own while you second guess every decision, you can sit back and relax while someone who has been in the trenches for over half a decade provides you working solutions and insights over a few hours.
You're getting a comprehensive learning experience that was distilled from years of analysis and modelling while building dozens of data science deliverables.
What are good other courses and other books for learning Bayesian thinking
You may get more out of this course if you've looked at Fundamentals of Bayesian Data Analysis, and I've done my best to add in introductory material.
Other good books include Statistical Rethinking
Other good books include Statistical Rethinking
Can I get an invoice for work?
Often customers want to use company training budget to pay. If you pay upfront, you can get an invoice from the billing page in your dashboard, which you can then expense back to your employer.
Does this course have an expiry date?
No, the price gets you lifetime access to the course, especially as I add more content. It's a one-time-fee.
Why should I buy this course over a book?
While I agree you can learn from some of the excellent books out there. This showcases me talking through different examples, which is much like pair programming. It's based on real-life experience of teaching this to over 400 Data Scientists, and you won't find material like this anywhere else. I think screencasts are an excellent way to level up your skills.
How many hours of video is in this course?
There's over 5 hours of professionally edited video in this course.
Including the first ever screencasts of new tools such as Arviz and Rainier in any course.
Including the first ever screencasts of new tools such as Arviz and Rainier in any course.
With Theano being discontinued is it worth learning PyMC3?
We'll continue to support Theano for the next few years as part of the PyMC3 project, so you can consider this long term supported for the next few years.
Future work in PyMC4 will involve a different API but similar concepts. The current plan is for PyMC4 to be built on top of Tensorflow.
See the announcement here
Future work in PyMC4 will involve a different API but similar concepts. The current plan is for PyMC4 to be built on top of Tensorflow.
See the announcement here
What is the target audience with this course?
My target audience is people with some statistics background or machine learning background. If you know some Python (or another suitable language like Scala, R, Java) and some Machine Learning you should be fine.
My aim is to present things in a 'hacker' friendly way.
My aim is to present things in a 'hacker' friendly way.