State Space model estimation using STAN as the inference engine. These models are usually used for time series analysis, and here we apply it to analyzing the incremental sales from a drinkware company's marketing campaign. This section outlines the theory of the model and the next will have the actual applications.
Polychoric correlation using a bayesian framework in stan. We use likert data, the type associated with "strongly agree" statements found on a survey or questionaire.
Bayesian poisson regression with a Metropolis Hastings sampler. We use poisson regression to model fish caught in different ponds. Our model is eventually used to determine what pond is the best to fish at next, given a choice between two new ponds. Along the way, we run some MCMC diagnostics and go into how well our chains converged.
Who doesn't like looking at moving disitributions? Whether you're setting up an AB test or checking your how strong your priors need to be, this tool will most definitely help!
Every wanted to play rock paper scissors wirelessly with a 3D printed hand? Well I made that dream a reality. Check it out!
Just a simple webapp showing TF JS functionality for building simple models in the browser. The example here builds a single layer hidden feed-forward neural network all in your browser. Make sure to try out some different functions!