search

uchicago-computation-workshop / 2018_spring_conference

0 stars 1 forks source link

Erin M. Ochoa #9

Open emochoa opened 6 years ago

AlexanderTyan commented 6 years ago

Very good graphics and intuitive and informative graphic transitions!

jamesallenevans commented 6 years ago

SKATER method region data climate conditions (weather stations) cool classification approach but initially too general/coarse 20 climate regions violent crime as a function of weather similarity

rickecon commented 6 years ago

Solid presentation. I didn't love the pink highlights (hard to see when the background was light). But it was definitely distinctive. The content was very good. Great balance of text, equations, figures. And you spoke words that were not verbatim from the slides. Nice.

I'm not sure about that definition of optimal regions.

bethbailey commented 6 years ago

Interesting topic, and very clear presentation! You clearly described your methods, which made it easy to follow. Good luck your analysis of crime.

bensoltoff commented 6 years ago

I remember this problem stymying you on the paper last year. Glad to see progress on this front.

Your approach sounds similar to selecting k number of topics in an LDA model for text data. That uses perplexity scores that are also monotonically decreasing. I wonder if anyone has developed an approach similar to your proportion measure. If so, it might work for your problem as well (or validate your current procedure)

xywu-soc commented 6 years ago

Interesting topic and solid work!!

banerjeeesha commented 6 years ago

Good thorough work!

bnroths commented 6 years ago

Awesome work and well explained. I was wondering what software/package you used to create the Thiessen map, I'd like to check it out