Lately, we have made a lot of progress on the BayesHive web application:
We have made huge usability improvements that combine to make the overall experience of using BayesHive more enjoyable. Like many other web applications, we are closing the gap between the desktop and browser. These improvements have particularly occurred in the document editor, where we now have syntax highlighting, persistent undo, online syntax and type checking and an editing area that scales to the browser size.
In the dynamical systems builder, you can now forecast the evolution of the dynamical system based on the historically observed data. We'll do a blog post on how this works.
There are still some gaps in our documentation, but our reference manual is becoming a usable document. We are working on additional tutorials.
We have been improving the layout of the documents produced using bayeshive. We now have margin notes1 which can also contain equations2 and small plots. You will see these in the tutorials we are writing.
Taken together, these improvements mean that BayesHive is just about ready for wider adoption. There are will still be bugs around, but we should be able to fix those quickly when we find them.
For the near future, here are some of the things we are working on right now:
We are working on more builders for statistical models. These will let you fit simple probability distributions and autoregressive timeseries models using the same point-and-click interface that you get for the regression, ANOVA and dynamical systems builders.
We want to improve the social aspects of BayesHive. In particular, this means being able to showcase your shared items (documents, datasets and models) to users who do not have a BayesHive account or are simply not logged in. We will create a web destination where you can browse, search and play with shared items.
We are also still working on a new inference engine that will allow you to swap and interleave different MCMC strategies, as well as writing new strategies yourself.