Marco Z
Marco Z
Possibly a binary in the app/ folder with an end-to-end workflow. Then we can split back anything good that comes out of this into the main library
Medium-long term : the loading/parsing machinery is growing in size and scope (see #22 , #29 ), so those functions and types could be gathered in a separate `datasets-core` package....
The Netflix Prize dataset uses a custom parser because one data example does not fit into a single dataset row (such as CSV data) but has a custom "stanza-based" format....
Some references : * StatsBase.jl http://juliastats.github.io/StatsBase.jl/stable/index.html
I've started adding some code from my `decision-trees` project under the `Core.Numeric.Statistics` and `Core.Data` namespaces . There is some machinery that could be re-used (for example the `Dataset` abstraction for...
~~As sub-project~~ Better to move parts of the backend and the typeclass interface for now
Currently the homepage features a rather uninspiring wall of text, much of which could be moved to the various sub-pages listed in the menu bar. I think we could take...
The Hypothes.is service relies on annotatorJS (http://annotatorjs.org/). The former is BSD licensed whereas the latter can be licensed under the MIT. Hypothesis may be embedded as a call to remote...
I think there are some false positives in the node detection algorithm; Scala keywords such as `as`, `is`, `type`, `containing` are output as isolated nodes. Is this intended behaviour? I...