Shinyverse - A multi-app shiny app
shinyverse is a Shiny web application that works as a compilation of some of shiny apps I have created. Currently it contains three applications:
- Adaptive Rejection Sampling
- Poisson with change point
- Normality tests
Poisson with change point
In this game it is possible to compete against Gibbs Sampling methodology. The game is very simple: It is a Poisson model with an unknown change point and the main aim is to guess where the change point is.
First, the user can configure the MCMC data and specifications. Then you can click on the graph to try to find the change point. After that, you will see where the Gibbs Sampling estimate is, as well as some statistics about the game.
Adaptive Rejection Sampling
Adaptive Rejection Sampling (ARS) is a blackbox methodology used to sampling from unknown log-concave distributions. It is also used together with a Metropolis Hastings step to sampling from unknown distributions during MCMC (called in this case ARMS).
In this application the user can understand how the Adaptive Rejection Sampling works in practice. As inputs the user can select some log-concave distributions and setup parameters and limits as well number of observations to simulate.
As outputs the user have 4 plots and a sliderInput. For each iteration selected we have:
- Log-function: You can see the points used to create the lower and upper hulls
- Histogram: Data sampled
- Function: You can see the points used to create the lower and upper hulls in original scale
- Acceptance rate: Acceptance rate until the selected iteration
For more details the user will find a "help page" on the application.
In this application we can see by some transformations how the most famous tests of normality are executed. The user can obtain samples of a desired distribution and apply the desired transformation and see the performance of the normality tests.
I hope you enjoy!