🌙
Bayesian Methods Working Party
Aims
What Are We Modeling?
What Would Success Look Like? (Latent Track)
Concepts and Definitions
Linear Algebra
PCA Concepts
Bayesian Concepts
Bayes Theorem
Priors
MCMC
Notation
Formulae
Q & A
Classical PCA
1 Raw Spot Data
1A Checks
2 Rectangular Matrix
2A Checks
3 Logarithms
4 Difference
5 Demean
6 Covariance Matrix
7 Eigendecomposition
8 Variance Explained
9 Projecting Co-ordinates
10 Component Score Distributions
Bayesian PCA (Bolt-On)
1 Specify the Prior
2 Build the Probabilistic Model
3 Sample the Posterior (MCMC)
4 Convergence Diagnostics
5 Posterior Distribution of Parameters
6 Posterior Predictive Distribution
Bayesian PCA (Latent)
1 Specify the Prior
2 Build the Probabilistic Model
Conference
Slide 1
Slide 2
Slide 3
Slide 4
Conference Questions
Conference Info to Build In
Dump
Next Steps
Purpose of Slide 3
Draft Points
One of the advantages of Classical PCA is that it is easy to implement, communicate and understand. The extent to which this changes under the Bayesian model is ....
why do all of this?
Soundbites
