Chicago Machine Learning Meetup: Time Series Analysis
Details
Join us on July 17th for an interactive study session on time series analysis, facilitated by Ben Smith.
The talk will start with a discussion of what a time series is, including methods for storate and representation to facilitate analysis and learning. Classification and clustering are the most basic (and successful) "applications of similarity" and Ben will go over the most practical approaches with time series. He'll avoid proofs, since the most effective methods should be understandable by anyone with high school level math. For those with more math under their belt (or simply more curiosity) he can go into more depth depending on what people find interesting.
Resources:
http://www.cs.ucr.edu/~eamonn/SAX.htm (http://www.cs.ucr.edu/%7Eeamonn/SAX.htm)
http://www.cs.ucr.edu/~eamonn/iSAX/iSAX.html (http://www.cs.ucr.edu/%7Eeamonn/iSAX/iSAX.html)
Excellent starting points for a variety of time series information, specifically the publications and references further down the page. Professor Keogh's Google talk linked from the SAX page covers motif discovery (finding interesting time series).
tutorial (http://www.cs.ucr.edu/%7Eeamonn/iSAX/tutorial_ICDM06.ppt)
A very thorough tutorial covering the time series state of the art (really large set of slides).
Compressive sensing is the tip of the spear as far as new theory and research, but recommended only for the most curious and motivated.
Ben is a former navy cryptologic technician and intelligence data collection specialist. He has an MS in computer science from the University of California Riverside. His work included being the first to generate and store every reachable state of the Rubik's Cube (All 4.3x10^19 of them!). He is currently working as an independent contractor in real estate, attempting to bring valuation and investing out of the superstitious dark ages using machine learning.