en
0.25
0.5
0.75
1.25
1.5
1.75
2
Clustered Graph Randomization: Network Exposure to Multiple Universes
Published on Sep 27, 20135880 Views
A/B testing is a standard approach for evaluating the effect of online experiments; the goal is to estimate the `average treatment effect' of a new feature or condition by exposing a sample of the ove
Related categories
Chapter list
Clustered Graph Randomization: Network Exposure to Multiple Universes00:00
A/B testing a social product change - 100:09
A/B testing a social product change - 200:41
A/B testing a social product change - 301:18
A/B testing a social product change - 402:00
A/B testing a social product change - 502:12
A/B testing a social product change - 602:25
Measuring ‘Average Treatment Effect’03:22
Average Treatment Effect: what changes?05:43
Agenda07:27
Defining ‘network exposure’ - 109:21
Defining ‘network exposure’ - 209:23
Defining ‘network exposure’ - 311:10
Clustered graph randomization13:04
How to partition the graph? - 113:14
How to partition the graph? - 213:42
Cluster and randomize14:07
Cluster and randomize... finely.14:13
How to cluster? What algorithm?14:45
Probabilities under clustered coin flips - 116:51
Probabilities under clustered coin flips - 217:38
Characterize ATE Variance17:58
ATE Variance18:06
Cycle graph example - 118:33
Cycle graph example - 218:45
Cycle graph example - 319:16
Cycle graph example - 419:19
Cycle graph example - 519:37
Cycle graph example - 619:44
Cycle graph example - 719:47
Cycle graph example - 820:08
Restricted growth condition20:34
Restricted growth graphs - 120:35
Restricted growth graphs - 220:55
Restricted growth graphs - 321:38
Clustering restricted growth graphs22:11
Conclusions - 122:49
Conclusions - 222:50