yaledhlab / lab-workshops Goto Github PK

This is a request for a workshop on how to use Github :)

This is a request for a workshop on web scraping, to support Sam and others on campus.

I have used the machine learning workshop and remixed it into a software carpentry-style episode https://data-science-for-biotech.github.io/python-pandas-viz-ml/03b-machine-learning/index.html. I found your workshop very approachable and of great help. So thank you for making it available!

I just don't get Palladio. What would I use it for and how?

Several bookworm scripts (specifically, those for flushing and reloading data) assume 'dbadmin'@'localhost' rather than 'dbadmin'@'127.0.0.1'. Not sure how interchangeable they are; we could either add in localhost as an addition or swap it out.

Can't wait to build and experiment!

We're running a live instance here: http://bookworm.library.yale.edu/collections/vogue/

The latest code, more up-to-date code is here:

API: https://github.com/bookworm-project/bookwormDB
Default FrontEnd (HighCharts line graph): https://github.com/Bookworm-project/BookwormGUI

In addition, Ben has been working on d3-based visualizations from the same API backend, here: https://github.com/bmschmidt/BookwormD3

We would have texts ready to go for attendees. Ideas for texts to prepare:

• Vogue photo captions [PL]
• Medical Heritage Library pre-1923 texts [MG]

Do we know if the Ubuntu script is modifying /etc/mysql/my.cnf (that is, the central one, not ~/.my.cnf) to include the customizations that Bookworm uses? This may not have any effect on small datasets, but will probably lead to an underside TMP table error on larger ones...

What's left to do on the word-vectors workshop:

• fix PDF links
• Add Hierarchical explanation
• explain centroid more?
• Review for typos/mistakes
• Add resources section/links to more info

I just met with Dominique in the library to discuss her team's desire to visualize some linked open data with D3. I thought we should do a lab session on the topic, helping folks munge XML data as exported from Dominique's server (which is a large linked open data server) into the miserables.json format used by several d3 examples. We could then show participants that this same data can be visualized with:

a force-directed graph: https://bl.ocks.org/mbostock/4062045
a cooccurrence matrix: https://bost.ocks.org/mike/miserables/
an arc diagram: http://bl.ocks.org/enjoylife/4e435d329c2c743da33e

That could help open up discussion on the virtues of different data visualization techniques...

Discussing pull requests and merging with lab members

This is a request for a workshop on how to find overrepresented terms in a corpus. Emerged from a conversation with Clay on his work

I'm able to get Federalist mostly working, except. Except that I get the spinning loading wheel of death. In my apache error log, there's a mostly helpful error:

[Tue Aug 23 20:50:42.716682 2016] [cgi:error] [pid 12821] [client 130.132.173.70:43443] AH01215: ERROR:root:/etc/mysql/my.cnf, referer: http://54.152.238.88/federalist/

Don't try to access the instance right now, as it's shut down. Do I need my bookworm db credentials in /etc/mysql/my.cnf? I don't really care why it's looking there instead of ~/.my.cnf as long as I can make the connection, though it's a little odd.