Fork me on GitHub
#data-science
<
2017-05-31
>
mikera01:05:21

@otfrom there's no reason that reducers couldn't be made to work as arguments to c.c.m.dataset API functions. And then you could use transducers as usual. If there's a use case, happy to consider.

mikera01:05:30

The idea of clojure.core.matrix.dataset is similar to the main clojure.core.matrix API: we provide a set of useful and general purpose API functions that can be supported by multiple implementations (e.g. you can have data tables from spark, JDBC result sets, Clojure data structures etc.)

otfrom09:05:45

@mikera I'd never thought of using transducers on a c.c.m.dataset w/out changing it to a row-seq first. I'll have a look at the API again and see if I can figure out how to get a reducer in there (cc @elise_huard)