nasyring / gsoc-bayesbd

GSOC-BayesBD

To install the BayesBD package you will need:

1. R, version 3.4.1 was used to build BayesBD.
2. Rtools.
3. The devtools package.

Steps:

1. Open Rgui.
2. Install the "devtools" package if you have not already using the dropdown "Packages -> Install package(s)...". Load the devtools package using the command library(devtools).
3. Submit the command: install_github("nasyring/GSOC-BayesBD", subdir = "BayesBD").

Alternative Installation: Thie version 1.2 is built for Windows on CRAN. Install using the menu in Rgui.

Learning about the package:

1. Load the package with the command libary(BayesBD). Bring up documentation using the command ?BayesBD, ?fitBinImage, ?fitContImage.
2. Run the examples therein.
3. Run BayesBDshiny() to bring up an interactive, html-based version of the package.

gsoc-bayesbd's Issues

Performance comparison vs. the benchmark

Could we have some comparison in terms of the speed gain by doing the cpp, compared to the original package? Both functions should produce same results when the random number generators are controlled.

Shiny Integration

I have posted a very simple example (BayesBD_shinybinary) of using Shiny to interface with BayesBD. This is meant to start a conversation about how Shiny might be useful for sharing this work with others and facilitating visualization.

Currently, the app is barebones. It is set up to show examples of binary image boundary detection and allows the user to specify input values over a range for model parameters and sampling settings. It uses c++ code and keeps the number of samples low so that it updates quickly. Since this is a work in progress, it is not yet posted to Shiny.io, the free server available from RStudio which allows you to actually post Shiny apps to the web. For now, it just runs off of the local machine.

Update to the latest version

Please update all the files and documentation to align with the latest R package in CRAN.

Code blocks & R crashing

In order to debug the R GUI crash issue, will it help if break the code into blocks and test them separately? Also, there are a lot of Rcpp's in the draft code, indicating pretty intense calls of C++ from R, does it seem to be a reason for the crash?

Code Documentation & R package

I guess it's now also a good time to document the code and create the updated R package. It usually takes a while to put things into a 'final' product so better to start slightly earlier.