Vishal Patel | A Practical Guide to Dimensionality Reduction Techniques

[emmcauley]

Video URL: https://www.youtube.com/watch?v=ioXKxulmwVQ

Contents

0:03 Speaker Introduction 0:20 Presentation Agenda 1:24 Dimensionality Reduction (DR), a brief overview 3:30 DR techniques are used as part of data preparation 4:50 Overview of the Springleaf data used throughout the presentation 5:51 Downsides of the “Kitchen Sink” approach of using the entire data set 8:57 Different DR techniques 10:02 Percent Missing Values 12:50 Amount of Variation (AOV) 13:47 Pairwise correlation 17:37 Multicollinearity 20:14 Principal Component Analysis (PCA) 23:46 Cluster analysis 26:19 Correlation with the target 27:17 Forward/Backward/Stepwise selection 30:58 LASSO 32:01 Tree-based 32:41 Proposed workflow of some DR techniques 33:09 AUC comparison of some DR techniques on Springleaf data 34:11 Summary of DR goals 34:53 Thank you; contact information 35:13 Why would you see dropping of an important feature in iterative stepwise selection? 36:43 What is the difference between LASSO and Logit (stepwise + not) on the AUC curve plot? 38:45 How sensitive is the order of the workflow?

[McArthurAdal]

i wonder why the timestamps haven't been applied.