CS 277: Data Mining

Spring 2013

Instructor: David Newman

Lectures: 12:30 to 1:50pm, Tuesdays and Thursdays, DBH 1500

Office Hours: Mondays, 11:00am - noon, DBH 4064

Reader: Sridevi Maharaj, sridevi.m@uci.edu

Message Board: https://eee.uci.edu/toolbox/messageboard/m13971/f41173/

Software: information about MATLAB

Background Reading : additional papers to supplement the text

Project Guidelines, including due dates, instructions, and links to data sets

Schedule

New

Homework 2 (due in class May 21) *** Please ignore 2010 due date in pdf!
Homework 2 info
Homework 2 first article to read
Homework 2 second article to read
Homework 2 third article to read

Syllabus

(follows CS 277 Syllabus from Prof. Smyth. Below syllabus subject to change.)

Introduction to Data Mining:

Topics
- basic concepts in data mining
- data measurement
- exploratory data analysis
- data visualization
Reading
- Mining the past to determine the future: problems and possibilities, D. J. Hand, International Journal of Forecasting, 25(3), 441-451, 2009.
Links
- UCI Machine Learning Repository of research data sets
- KD nuggets (industry-oriented Web site for news related to data mining)
- Commentary in Science, Dec 2009, on privacy in data mining
Slides
- Introduction to Data Mining [PPT] [PDF]
- Measurement and Data [PPT] [PDF]
- Exploratory Data Analysis and Visualization [PPT] [PDF]
Homeworks
- Homework 1
  - Data and MATLAB files for homework 1
- Homework 2
  - data and additional information
  - summarize.m (template)
  - define_title_strings.m

Basic Principles of Data Mining

Topics
- predictive modeling: classification and regression
- model fitting as optimization
- evaluation of predictive performance
- overfitting, regularization
- other data mining tasks: clustering amd pattern detection
Reading
- other
Links
- Lasso method for regularized regression
Slides
- Regression [PPT] [PDF]
- Classification [PPT] [PDF]

Text Mining

Topics
- information retrieval and search
- text classification
- unsupervised learning
Reading
- Text classification using naive Bayes from Manning, Raghavan, and Schutze
- SVMs with applications to text classification from Manning, Raghavan, and Schutze
- Tutorial papers on topic modeling:
  - Steyvers and Griffiths
  - Blei and Lafferty
Links
- Mark Steyvers' MATLAB code for topic modeling
- Dave Blei's page on topic modeling (C code, demos, and mailing list)
Slides
- text classification [PPT] [PDF]
- text mining and topic models [PPT] [PDF ]
- notes on graphical models [PPT] [PDF]

Recommender Systems

Topics
- recommender data, Netflix prize data
- nearest neighbor algorithms
- matrix decomposition algorithms
- efficient algorithms for large data sets
- modeling systematic effects
Reading
- chapter in Web book
- Matrix factorization techniques for recommender systems (2009 IEEE Computer paper by Koren et al.)
- Tutorial on collaborative filtering recommender systems by Schafer et al, 2007
- Recommender algorithms used by Amazon.com (2003, IEEE Internet Computing article)
- Evaluating recommender systems (2004 ACM TOIS article)
Links
- Proceedings of 2009 ACM Conference on Recommender Systems (RecSys 09)
Slides
- Recommender systems [PPT] [PDF]
- Netflix case study [PPT] [PDF]

Web Data Analysis

Topics
- Web data: collection and interpretation
- analyzing user browsing behavior
- learning from clickthrough data
- predictive modeling and online advertising
- link analysis and the PageRank algorithm
Reading
- Web search basics, chapter 19 from Introduction to Information Retrieval by Manning, Raghavan, and Schutze
- PageRank and related algorithms
  - Anatomy of a large-scale hypertextual Web search engine, original paper from 1998 by Brin and Page on PageRank
  - Deeper inside PageRank, Langville and Meyer, 2004
  - A Survey of Eigenvector Methods for Web Information Retrieval, Langville and Meyer, 2005, SIAM Review
  - Link analysis for the Web graph, chapter 21 from Introduction to Information Retrieval by Manning, Raghavan, and Schutze
- Cadez et al paper on clustering user browsing sequences with mixtures of Markov chains
- Joachims paper on learning from clickthrough data
Links
- WWW 2009 Conference and Proceedings
- Stanford class on "computational advertising"
Slides
- Web link analysis [PPT] [PDF]
- Web usage mining [PPT] [PDF]

Social Network Analysis

Topics
- descriptive analysis of social networks
- network embedding and latent space models
- network data over time: dynamics and event-based networks
- link prediction
Reading
- overview of network analysis methods (e.g., Newman et al)
- overview of SNA concepts
- review paper on community detection in networks
- specific papers by Watts, Leskovecs, Barabasi, etc
Links
- Class on network data analysis by Jon Kleinberg
- slides from Kleinberg invited talk on network algorithms
- International Network for Social Network Analysis
Slides

Tentative Schedule

Week 1:
- Tue Apr 2: Read this Pedro Domingos paper
- Thu Apr 4:
Week 2:
- Tue Apr 9:
- Thu Apr 11: Project proposals due in class
Week 3:
- Tue Apr 16:
- Thu Apr 18: HW1 due in class
Week 4:
- Tue Apr 23:
- Thu Apr 25:
Week 5:
- Tue Apr 30: Project progress presentation in class
- Thu May 2: Project progress presentation in class
Week 6:
- Tue May 7:
- Thu May 9:
Week 7:
- Mon May 13: Office hours cancelled
- Tue May 14: ***NO LECTURE*** Continue work on project
- Thu May 16:
Week 8:
- Tue May 21: Scientific writing. HW2 DUE IN CLASS
- Thu May 23: Scientific writing
Week 9:
- Tue May 28: Project presentations in class
- Thu May 30: Project presentations in class
Week 10:
- Tue June 4: Project presentations in class
- Thu June 6: Project presentations in class; Final project report due in class

Grading

Your class grade will be based on a class project (70%) and two homeworks (total of 30%, each HW is worth 15%). The projects will require submission of a progress report during the quarter, a presentation in class during Weeks 9 or 10, and a final report.

Academic Honesty

It is the responsibility of each student to be familiar with the UCI Senate Academic Honesty Policies. For homework assignments and projects you are allowed to discuss ideas and concepts verbally with other class members, but you are not allowed to look at or copy anyone else's written solutions or code relating to homework assignments or projects. All material submitted must be material you have personally written during this quarter. Failure to adhere to this policy can result in a student receiving a failing grade in the class.