Statistical Change Detection for Multi-Dimensional Data  thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Statistical Change Detection for Multi-Dimensional Data

Published on Aug 15, 20077797 Views

This paper deals with detecting change of distribution in multi-dimensional data sets. For a given baseline data set and a set of newly observed data points, we define a statistical test called the de

Related categories

Chapter list

Statistical Change Detection for Multi-Dimensional Data00:03
Motivation Example: Antibiotic Resistance Pattern00:25
Problem Definition01:27
Related Work01:45
Hypothesis Test Framework02:41
Density Test High-Level Overview03:14
Step 1: Kernel Density Estimate (KDE)04:11
Choose Bandwidth by MLE/EM05:44
Effectiveness of EM Bandwidth07:13
Step 2: Define and Calculate 07:49
Step 3: Derive the Null Distribution08:52
Estimating 10:02
Step 4: Calculate Critical Value and Make a Decision 11:13
Density Test – All 4 Steps12:28
Run Density Test in 2 Directions12:38
False Positive13:03
False Negative on Low-D Group13:54
False Negative on High-D Group14:19
Scalability14:30
Conclusion15:17
Thanks15:44