You might also find the following useful or interesting, this is an
edited entry from my personal notebook from May.

* ProjectTemplate should be an opinionated framework.

* ProjectTemplate should commit to a software paradigm that can inform
architectural decisions and a statistical methodology that can inform
"opinionated decisions."

* ProjectTemplate should motivate convention over configuration.

* ProjectTemplate should include an object-relational mapper that
adopts the data mapper pattern rather than the active record pattern.

i.e. SQLAlchemy rather than ActiveRecord.

* ProjectTemplate should exercise a user experience that emphasizes
statistical knowledge rather than software knowledge.

e.g. Is the `diagnostics` directory meant for diagnosing programatic
bugs or model validation?

* ProjectTemplate should simplify routine chores.

e.g. How can we simplify the usage of Weka from R? Hadoop?

* ProjectTemplate should include preprepared scaffolding for routine
statistical procedures.

e.g. How can I easily create a new project for natural language
processing? Time series analysis? Geospatial analysis?

* ProjectTemplate should have temptable progressive-agnostic reporting.

e.g. How can I use Rook? WebSockets? brew?

* ProjectTemplate should broaden the scope of its importing mechanism
to support import from REST.

e.g. How can I import data from Twitter's OAuth-required API?


Tuesday at 5 works. You should stop by our office in DUMBO (18 Bridge
St., #4F), then we'll get a drink somewhere in the neighborhood.

I forgot to add my typical directory structure to my last e-mail. You
might find it interesting. I believe it's somewhat self-explanatory.

`data/`

`data/db/`

`doc/`

`log/'

`tmp/`

`tmp/cache`

`config/`

`lib/`

`report/`

`report/print`

`report/web`

`report/images`

`research/`

`research/clean`

`research/transform`

`research/model`

`test/`

`test/functional`

`test/performance`

`test/unit`
