Having seen demos using reflection/code introspection à la smalltalk in statistical programs pertaining to codebase analysis,
I feel there ought to be many interesting applications for statistical tooling, but I wasn't really able to think of any besides the obvious ones:
* interactive graphs with one-click link to underlying data
* visualize the program itself (i.e. the analysis plan) as an interactive graph (https://www.youtube.com/watch?v=baxtyeFVn3w&t=26m50s)
* for each result, one-click link to every element used in computation (= callees), and to every further computation using said result (= callers), essentially collapsing indirection (https://www.youtube.com/watch?v=baxtyeFVn3w&t=30m20s).
Reflection adds overhead, but for people doing small and irregular data (of which there are many) it seems to have potential. I would be interested in your opinion and ideas for additional applications of those principles.
Having seen demos using reflection/code introspection à la smalltalk in statistical programs pertaining to codebase analysis, I feel there ought to be many interesting applications for statistical tooling, but I wasn't really able to think of any besides the obvious ones:
* interactive graphs with one-click link to underlying data * visualize the program itself (i.e. the analysis plan) as an interactive graph (https://www.youtube.com/watch?v=baxtyeFVn3w&t=26m50s) * for each result, one-click link to every element used in computation (= callees), and to every further computation using said result (= callers), essentially collapsing indirection (https://www.youtube.com/watch?v=baxtyeFVn3w&t=30m20s).
Reflection adds overhead, but for people doing small and irregular data (of which there are many) it seems to have potential. I would be interested in your opinion and ideas for additional applications of those principles.