Feb 2, 2013, 5:32:46 PM2/2/13
to Jana Beck, mede...@googlegroups.com, Melissa Thielen, Bernard Farrell, Damon Muma, Aleksander Rozman
Those sound like the right steps, I can help with both, although I'm not sure if my R installation is up to date.
I don't know when I'll be able to try running R and python; if you submit some gists with errors, I might be able to help/explain things.
For d3, I intend to get fork some of mike's boxplots and heatmaps to try and port your charts in gist/d3 soon.
seems to be down for all examples (not just mine) ATM, as Bernard also noted.
It's not mine, it's a google app run my Mike Bostock. It simply re-serves a gist as a pre-formatted web page. This idiom (of re-using git-repos for other purposes, eg as the data-package for an application) is increasingly commone, ruby, node, python, et al all have package managers that intuitively deal with git-hosted repos for package management. (I'm curious if we can create a git-phr, to manage medical records, independently of the application consuming it, and to audit what effects software has on those records. With a git-phr, you could use a tool to compile the kind of report Jana created, and keep it around with your data, even if the app you are currently using doesn't know anything about it. Forever.)
We do need a visualization gallery app, similar to bl.ocks.org
, but focused on diabetic visualizations. Bernard, I commend you for setting up hosting stuff. I think github will serve most of our communication/infrastructure needs, but if you want to host bl.ocks.org
for us so we at least host visualizations composed in gist or similar reliably, that might be compelling. Or if someone wanted to come up with a gh-pages template for these visualizations, I'll port mine over and try to stub ut some more demos, and we can simply use github for this as well.
Bernard, we will be using the materials you've collected under the diabetes wiki heavily, I suspect. :-)
On Sat, Feb 2, 2013 at 1:28 PM, Jana Beck <jana.el...@gmail.com>
Great! Melissa, nice to meet you! (I just followed you on Twitter too, although I'm barely on there.)
D3 is definitely my next step, and it's good to know that there are resources for help on that, as I am new to it. (BTW, I got errors on both of your demos at bl.ocks.org
, don't know if that's their servers or something else.)
I've also started experimenting with hooking into R via Python here
. My first effort was mostly unsuccessful (it does produce the boxplot, but efforts to hook into the ggplot2 package failed), but I talked to some of my academic colleagues who use R a lot more than I do, and they managed to dig up some more extensive sample code than what's in the RPy2 docs, so trying again is on my list of things to do.
On Thu, Jan 31, 2013 at 6:16 PM, Benjamin West <bew...@gmail.com>
Jana, your demoes are very nice. Many of this is doable using d3.js
Have you seen the horizon charts advocated by Mike Bostock? I keep thinking they might come in handy.
On Thu, Jan 31, 2013 at 3:09 PM, Benjamin West <bew...@gmail.com>
The two of you may have overlapping skills that work well together. Surprisingly few people know python.
Here's a rough outline of all my stuff:
Insulaudit is for auditing lots of devices,
the decoding-foo are each to decode that very specific type of device.
I've found gist/blocks/d3.js very useful for experimenting with d3.
Backend, wise I was thinking of having lots of daemons, communicating using json, and continuing to experiment feature by feature before attempting to tie together any monolithic web-app.
Does working altogether under the "medevice" org umbrella make sense?
On Thu, Jan 31, 2013 at 2:46 PM, Benjamin West <bew...@gmail.com>
I'm reading through your github materials. I look forward to working together!
On Thu, Jan 31, 2013 at 1:22 PM, Jana Beck <jana.el...@gmail.com>
Amy Tenderich just sent me a heads up about the Diabetes Mine piece on you today, and I was very glad to see it. I was one of the patient representatives at this year's DiabetesMine Innovation Summit, and I've been doing some hacking of my own diabetes data. My main interest is in analysis and visualization: you can see some of the charting I've done with my Dexcom data here
I'm also working on creating an interactive visual logbook with my Dexcom data, insulin doses, meals, exercise etc. That's all in a GitHub project
here, with a bit of a (messy) demo here
I'm following a couple of your repositories on GitHub now. Forgive me if it's on GitHub and I missed it, but is there any chance you'll be hacking the Dexcom? I am dying to get away from their Windows-only software to get my data out via their export-as-csv option.
I'm also sending you an invite to the Diabetes Hacks and Data Vis Google group I created after the DiabetesMine Summit.