Hic sunt leones
Latin phrase reported on many maps indicating Terra incognita, unexplored or harsh land.
Dataframes in Clojure. Through pandas. On Python.
This is very alpha, things will change fast, will break and the API is neither complete, nor settled. Since a few people have started playing with this there's a Clojars project available. Please give feedback if you're using this, every kind of contribution is appreciated (for more info check the Contributing section). At the moment everything is mostly undocumented and untested, I'm currently adding them.
Panthera uses the great libpython-clj as a backend to access Python and get pandas and numpy functionality.
To get started you need python, pandas and numpy (the latter comes with the former) on your path. Usually a:
apt-get install libpython3.6-dev
pip3 install numpy pandas xlrd # the latter is for Excel files, if you don't care you can do withoutAfter this you can start playing around with panthera
(require '[panthera.panthera :as pt])
(-> (pt/read-csv "mycsv.csv")
(pt/subset-cols "Col1" "Col2" "Col3")
pt/median)The above chain will read your csv file as a DataFrame, select only the given columns and then return a Series with the median of each column.
panthera.panthera is the home of the main API, and you can find everything there. The advice is to never :use or :refer :all the namespace because there are some functions named as core Clojure functions such as mod which in this case does the same thing as the core one, but in this case it is vectorized and it works only if the first argument is a Python object.
All of the main numpy is wrapped and accessible through a single interface from panthera.numpy.
(require '[panthera.numpy :refer [npy]])
(npy :power {:args [[1 2 3] 3]})
;=> [1 8 27]
(npy :power)
; This arity returns the actual numpy object that can be passed around to other functions as an argumentFor every function there is a key, to check everything that is available just call the zero-arity version - (npy) - and you'll get a list of keys. To see how they work either check the official docs online or call (npy :your-key {:doc true}) to check the original Python docstring.
This is because while panthera.panthera is carefully wrapped method by method, but it's so large that at the moment just 35%-40% of its functionality is covered, I wanted all of numpy available, so the wrapper is fully automatically generated.
Numpy submodules (like random and linalg) will follow soon.
Please let me know about any issues, quirks, ideas or even just to say that you're doing something cool with this! I accept issues, PRs or direct messages (you can find me also on https://clojurians.slack.com and on https://clojurians.zulipchat.com).
You can find some examples in the examples folder. At the moment that's the best way to start with panthera.
Pandas is derived from "panel data" and somehow is supposed to mean "Python data analysis library" as well. Though it shouldn't have nothing to do with the cute Chinese bears, there are logos showing a bear.
Panthera doesn't pretend to be a clever wordplay because it doesn't need to. First off panthera is latin and it literally means "large cat", second though pandas are surely cute, pantherae are way cooler (and snow leopards also happen to be among the very few predators of pandas, but that's just a case...).
Copyright © 2019 Alan Marazzi
This program and the accompanying materials are made available under the terms of the Eclipse Public License 2.0 which is available at http://www.eclipse.org/legal/epl-2.0.