Data Science

Invariants

Saramaki et al. (2014): "Persistence of social signatures in human communication" link

Mobile Phone Data

Tsapeli and Musolesi (2015): "Investigating causality in human behavior from smartphone sensor data: a quasi-experimental approach" link
summary
This paper claims to present a generic framework for causal analysis using mobile phone data. A solution to that problem would obviously be a huge step forward, but unfortunately as far as I can tell what they are doing is essentially the same as a typical regression analysis using a kitchen sink of variables to try to remove confounding effects. What I would have liked to see was something like a generic instrumental variable or generic discontinuity design or something along those lines. More work in this area seems to be needed.

Predicting Things

Kosinski et al. (2013): "Private traits and attributes are predictable from digital records of human behavior" link
Peter M Krafft Last modified: Thu Jan 23 10:15:46 EST 2014