of difference between questions asked in machine learning and social science • great research: “Julian Assange is an emergent property of our model” • great discussion of privacy and ethics concerns faced by computational social science research
https://youtu.be/OK6M4w7LYIc?list=PLYx7XA2nY5Gf37zYZMw6OqGFRPjB1jCy 6 • http://docs.mybinder.org/faq • One click lets you instantiate all the code + Jupyter notebooks + raw data associated with some project • Makes reproducibility painless. Just add water Docker
Psychometrics for Data Quality Evaluation”, Katie Malone • slides here: https://civisanalytics.app.box.com/s/ur5pb5h5a0x2booouop1zazca4ef8r3k • Basic idea: • Use math behind Item Response Theory from social science to determine whether it’s worth it to invest in more data collection • How “smart” does your data seem when you “test” it with a bunch of different models (e.g. a 100 different random forests)? • Your data scores in a high percentile after answering all these “test questions” worth it to collect more data like this • Note all the smart ideas here are Katie Malone and the bad explanation is mine
minutes • You can watch all the presentations yourself! • Including tutorials! • https://www.youtube.co m/playlist?list=PLYx7XA2 nY5Gf37zYZMw6OqGFRP jB1jCy6 • Here’s a picture of a cat!