Fork me on GitHub
#data-science
<
2020-02-24
>
otfrom13:02:37

is cljplot still my best bet for creating charts by batch without going to python or R?

mkvlr13:02:52

plotly.js or vega lite Iā€™d say if your target is web

otfrom13:02:07

my target is pdf really otherwise I'd probably go with Oz for now

otfrom13:02:46

and standalong png files (as my users can't handle svg)

otfrom13:02:18

though I could batch up converting svg to png using batik (I've sketched that out before and used some clojure wrappers that will do it).

otfrom13:02:38

what I don't want to do is click save 10s to 100s of times of a generated web page

jsa-aerial16:02:01

As announced a while back, Saite now supports saving all the images as png with a single click. You can create (in principle) any number of plots/charts across one or any number of tabs and save them all with a single save request.

šŸ‘ 4
genmeblog08:02:08

@U0525KG62 take a look at clojisr (R). Works perfectly well and setup is easy. Though I think I will be back to working on cljplot next month.

otfrom09:02:02

@U1EP3BZ3Q I'll be doing some work with cljplot for the next week or two, at least as a user. I'm not sure I'll be doing anything I haven't done previously. I'll still have to use R (or shell out to javascript) for the sankey charts I need.

metasoarous17:02:56

@U0525KG62 You can always take the vega and use the npm vega-cli tools for now.

metasoarous17:02:01

I've been on vacation and so not spending a ton of time on Oz last couple of weeks, but next month this is one of the top things on my list of priorities.

metasoarous17:02:54

First step will be support for calling out directly to the vega-cli tools from Oz when installed. I have a few ideas for how to avoid having to do that, but will need a bit of tinker time to sort out which is best.

metasoarous17:02:55

There are some nontrivial design decisions in how I want to expose some of these features, API wise, which is part of what has kept me from doing this already (in addition to some of the technical challenges). But this is high priority for me, as it unlocks a number of other issues I have planned to tackle, so please stay tuned!

ā¤ļø 4
Daniel Slutsky16:02:44

Updated the list of libraries at the scicloj website: https://scicloj.github.io/pages/libraries/ Any comments? Anything missing? šŸ™

andy.fingerhut18:02:21

I've asked a question on the #clojure channel about a time/space efficiency tradeoff for the data.csv library, but I would guess a similar tradeoff exists for some other libraries for reading large data files in multiple different formats. I would guess that some libraries for reading large data files that return numeric type values rather than strings might not have that issue -- it may be specific to strings in the returned data.

šŸ‘ 4