I have not tried to use core.matrix.dataset directly as it feels too low level with respect to the convenience routines you get in pandas (in this sense, core.matrix.dataset is more akin to numpy in Python). I have used Incanter’s dataframe-like functionality and found it very helpful if you are used to pandas and R. And, it looks like Incanter is moving to core.matrix under the hood. https://data-sorcery.org/category/incanter/

I think the functionality noted in this thread need good examples in Clojure…

>>> + joining (inner / outer / left) + filtering by rows and columns + time series offsets + row and column unions, including with ordered time series + rolling transformations + group-by + integration with automatic datetime parsing in xts (e.g. xts[‘2014::2015’] selects all items in years 2014 and 2015)

https://groups.google.com/forum/#!topic/numerical-clojure/MxvnITBUvmQ

I see that as wrong - numpy is more like core. matrix

this may be relevant https://github.com/sbelak/huri

he implemented some of those (joins, etc) but for clojure collections

Incanter is pretty much at a standstill at this point unfortunately

https://github.com/cgrand/xforms