The Gender Divide of Austin Tech Meetup Speakers
The Austin Writeup:
Work in progress: (http://amaliebarras.github.io/meetupSpeakerVizs)
Help close the gap:
Coming soon: http://wespeaktoo.org/austin
In meetup's api console I grabbed Austin's top 50 meetup groups in the 'tech' category, sorted by members.
In order to grab the individual events for these meetups, I know I'd need just the
url-group-name for each. I converted the output json into a CSV using an online tool, and deduplicated on
Then, I went through and deleted groups that I knew were not speaker events (such as OpenHack Austin). Because there were still over 45 groups meeting this criteria, I chose a sample of these groups to visualize, sorting the groups by total member count, and keeping every third group.
I made this into a narrowed-down meetup group list, so that with a Meetup API key, and thanks to Dash's commit on 11/30 on I could use python to run meetupdatagrab.py and to create a doc with all the events, as well as a template to notate the events with the speaker gender.
** Note: ** Before you run python, you'll need to get setup.
- Grab your API Key from meetup.com
- Paste API key into meetupDataGrab.py
- Make sure you've got python install
brew install python
- This also installs pip. Use pip to install virtualenv, which will let you create a virtual environment
pip install virtualenv
- Inside of project directory, initialize a python environment
- Now activate the virtual environment
- This script uses pandas, so install that using pip
pip install pandas
- Run script using
- Lastly, deactivate the virtual environment
Everyone I've talked to about this project has questioned my choice to manually encode this data when there are pronouns and databases out there. The honest answer is that I thought that I could do it faster. For what it's worth, Dash did experiment with creating notationbot.py, but it wasn't able to account for multi-speaker events. The process is super nuanced too. Although in most cases the gender of the speaker was identifiable by the pronoun used in the event description, where a pronoun was not given, I searched for other online resources that did provide a pronoun. And in a few rare cases I used name to indicate gender. PLEASE let me know if you spot an issue, I'd love to have the most accurate data possible.
I ended up opening the template in google sheets and notating there.
This is absolutely the most painful part of this exercise. Another developer is working on a tool for future local champions who want to replicate this analysis.
When done, I saved a copy into the repository as techMeetupsNotatedDone.csv.
I used gh-pages to create a writeup site, including copies of Kate's d3 visualizations as iFrames, changing only the data URL within each of them.