0.25
0.5
0.75
1.25
1.5
1.75
2
On the Stratification of Multi-Label Data
Published on Oct 03, 20114325 Views
Strati ed sampling is a sampling method that takes into account the existence of disjoint groups within a population and produces samples where the proportion of these groups is maintained. In sing
Related categories
Chapter list
On the Stratification of Multi-Label Data00:00
Stratified Sampling00:09
Stratifying Multi-Label Data (1)01:08
Stratifying Multi-Label Data (2)01:44
Stratification Based on Labelsets (1)02:54
Stratification Based on Labelsets (2)03:46
Statistics of Multi-Label Data04:10
Iterative Stratification Algorithm05:41
Example (1)07:14
Example (2)07:49
Example (3)08:03
Example (4)08:16
Example (5)08:18
Example (6)08:30
Example (7)08:58
Example (8)08:59
Example (9)09:01
Example (10)09:01
Example (11)09:02
Example (12)09:03
Example (13)09:04
The Triggering Event09:12
Subsets Without Label Examples10:17
Comparison of the Approaches10:39
Experiments11:46
Distribution of Labels & Examples12:24
Labels Distribution (normalized)13:30
Examples Distribution14:39
Subsets Without Label Examples15:53
Variance of 10-fold CV Estimates17:03
Average Ranking for BR (1/3)17:30
Average Ranking for BR (2/3)18:09
Average Ranking for BR (3/3)18:23
Average Ranking for CLR18:40
BR vs CLR18:59
Conclusions19:39
ecmlpkdd2011_tsoumakas_stratification_01_Page_3720:27