Meng's demonstration of the impact of bias on the reliability of Big Data has to be one of the most impressive and important results in statistics for years.

If U.S. Census isn'f for inference purposes, is it just simply to be used as data when applicable? How does one take into effect the data collected?

34:00 does this bypass sampling error fatal reducing hundreds of millions to hundreds w/spme corr checks. GaWd I guess it depends. Nonrandom (but what is random). & no stratification, no time addedyet - rfm

54:30 mixed resolution cross tx cf

47:30 decisions from data

38:00 pop inference is not wa big data is for, indiv prediict =1...rite? Quality //qt

any thoughts on what this might mean for autonomous driving vehicles?

RSS is awesome