en-de
en-es
en-fr
en-sl
en
en-zh
0.25
0.5
0.75
1.25
1.5
1.75
2
Spooky Stuff in Metric Space
Published on Feb 25, 20076295 Views
Decision trees are intelligible, but do they perform well enough that you should use them? Have SVMs replaced neural nets, or are neural nets still best for regression, and SVMs best for classificatio
Related categories
Chapter list
Spooky Stuff Data Mining in Metric Space00:00
Motivation #100:51
Motivation #1: Pneumonia Risk Prediction01:06
Motivation #1: Many Learning Algorithms16:17
Motivation #2: SLAC B/Bbar16:48
Motivation #2: Improves Calibration Order of Magnitude21:56
Motivation #2: Significantly Improves SLQ22:38
Motivation #223:38
Motivation #323:46
Motivation #323:47
Scary Stuff24:08
Scary Stuff26:18
Scary Stuff27:37
In this work we compare nine commonly used performance metrics by applying data mining to the results of a massive empirical study35:49
10 Binary Classification Performance Metrics36:42
lift = 3.5 if mailings sent to 20% of the customers37:31
better performance38:05
Predicted 1 Predicted 038:58
ROC Plot and ROC Area39:21
diagonal line is random prediction40:21
Calibration Plot40:46
Base-Level Learning Methods41:03
Data Sets41:08
Massive Empirical Comparison41:12
COVTYPE: Calibration vs. Accuracy41:40
Multi Dimensional Scaling44:46
Scaling, Ranking, and Normalizing45:32
Multi Dimensional Scaling46:12
Multi Dimensional Scaling47:17
2-D Multi-Dimensional Scaling47:42
2-D Multi-Dimensional Scaling51:18
Adult Covertype Hyper-Spectral52:54
Correlation Analysis53:21
Rank Correlations53:31
Summary57:05
New Resources58:26
Future/Related Work58:39
Thank You.58:46