2-Day Course: Bayesian Networks 101
Learn about the principles of Bayesian networks and how to apply them for research and reasoning with the BayesiaLab software platform.
This newly-created "executive" course is a streamlined version of the successful 3-Day Introductory BayesiaLab Course, which we have been offering around the world for almost ten years.
We have designed this new 2-day course for researchers who are interested in using Bayesian networks, but for whom the exercise-focused 3-day course might be too ambitious. The 2-day program removes the pressure of in-class exercises and gives you more time to just listen and learn. Using this format, we can accommodate a larger number of participants and make the tuition more affordable.
That is not to say that this is a "lite" program in any way. Within two days, we have to cover a large amount of material to portray the many facets of the Bayesian network formalism, including a good amount of theory. However, our emphasis is on how this theory becomes immediately relevant for many real-world research tasks. Overall, this course's focus in on leveraging Bayesian networks for applied research, with the objective of solving problems that currently appear intractable.
Although the Bayesian network paradigm can be utilized with many tools, this course is not vendor-neutral. For all demos and exercises, we exclusively use the BayesiaLab software platform. Therefore, you have access to a full license of BayesiaLab Professional for a 30-day period starting on the first day of the the course. This gives you the opportunity to replicate all examples from the course at your leisure on your own computer.
Date and Location
April 14-15, 2015, 9:00am-5:00pm (both days)
Course Overview
- Probabilistic Reasoning
- Fallacy of the Transposed Conditional
- Bayesian Networks for Statistical/Observational Inference
- Unsupervised Structural Learning with BayesiaLab
- Discovering Structure in EEG Data
- Supervised Learning (Predictive Modeling) with BayesiaLab
- Cancer Classification Example (Golub et al., 1999)
- Variable Clustering
- Data Clustering
- Unsupervised Structural Learning with BayesiaLab
- Bayesian Networks for Causal Inference
- Directed Acyclic Graphs
- Causal Identification Criteria
- Causal Effect Estimation
- Simpson's Paradox Example
- Simulation and Optimization with BayesiaLab
What's required?
- Basic data manipulation skills, e.g. with Excel.
- Working knowledge of specifying and estimating linear models.
- No prior knowledge of Bayesian networks is required.
- No programming skills are required. You will use the graphical user interface of BayesiaLab for all exercises.
Who should attend?
Applied researchers, statisticians, biostatisticians, data scientists, ecologists, epidemiologists, econometricians, economists, policy analysts, social scientists, decision scientists, market researchers, knowledge managers, students and teachers in related fields.
About the Instructor
Stefan Conrady is the managing partner of Bayesia USA, the North American partner of France-based Bayesia S.A.S.
Within Bayesian network circles, Stefan is perhaps best known for being the lead author of a series of 16 tutorials, all about applying Bayesian networks and BayesiaLab for research and analytics.
Stefan studied Electrical Engineering in his hometown of Ulm, Germany, and has extensive international management experience in the fields of business/product strategy, consumer research, and analytics, all with leading car companies, including Mercedes-Benz, BMW, Rolls-Royce, Nissan, and Infiniti.
Throughout his assignments in North America, Europe, and Asia, Stefan gained first-hand experience of how Fortune 100 corporations perform research, and, how many fundamental questions still await a satisfactory solution. Bayesian networks were one of the tools Stefan first explored while heading the analytics & forecasting group at Nissan North America.