Course Expectations and Tentative Syllabus
CSC:470 Special Topics: Data Mining Fall 2005
Olney 200 TR 11:00am-12:15pm
Professor: Dr. Michael Redmond
330 Olney Hall (215) 951-1096
http://www.lasalle.edu/~redmond/teach/470
Office Hours: MW 12-12:50pm, MW 5-6pm; TuTh 2-2:50pm
And at other times by appointment. Also, by phone and e-mail.
Text:
Witten, I. H., and Frank, E. Data Mining; Practical Machine Learning Tools and Techniques with Java Implementations, Morgan Kaufmann, 2000, ISBN 1-55860-552-5
Course Description:
This course is an introduction to data mining, with an emphasis on applying machine learning techniques for data mining. Data Mining involves digging information or knowledge out of a mass of raw data. Time Magazine named “Data Miner” number 5 on its list of top 10 jobs for the new century. Some practical applications include credit risk analysis and database marketing.
Machine Learning involves the attempt to get computer programs to acquire skills that they were not specifically programmed for and/or to improve with experience. Popular methods include methods of learning decision trees and decision tables, learning rules, and “lazy learning” – case-based reasoning. We will look in detail at several learning methods and their variations and how the machine learning methods can be used for data mining, including which algorithms can be used productively for what tasks and what data.
Also emphasized will be data preparation, and evaluation of results. The course work will involve some programming and will involve experimenting with public domain versions of famous machine learning/data mining programs to see how they work. Students will carry out a project with available data (probably using the public domain programs). This course counts for CSC and IT Elective credits.
Grading:
Assignments (5) 25%
Midterm Exam 20%
Project 25%
Presentation 5%
Final Exam 25%
Final Grades:
A 92-100 A- 90-91
B+ 88-89 B 82-87 B- 80-81
C+ 78-79 C 72-77 C- 70-71
D+ 68-69 D 60-67 F < 60
No make up exams unless arranged in advance. Make ups may involve double-counting of the final exam. Final exam is cumulative, but will focus more heavily on the (previously untested) final half of the course.
The assignments will mainly involve specific tasks experimenting with some famous machine learning data mining programs. The programs we will use will be public domain copies that are available over the Internet. Tasks will include data preparation, experimentation, and analysis of results. There will be one assignment involving writing a program – possibly implementing a well known machine learning algorithm and applying it to some existing data. Details of assignments will be presented as the semester proceeds.
The project may be done individually or in pairs. The project and presentation are related. Students will pick out some data that is of interest to them (personal or professional). They will choose a Data Mining goal with respect to the data. They will prepare the data, experiment with it, and determine results. The presentation will discuss the task and goals, an analysis of usefulness of available methods for the task, a summary of results and a conclusion. If the data is proprietary, be sure not to reveal proprietary aspects.
Materials: You may need at least 2 diskettes (or alternative media, such as CD-RW (the Olney 200 lab DOES NOT have zip drives)). You may need access to Java and the WEKA data mining software outside of class. It will be installed in Olney 200 and 200A, and may be downloaded for free from:
http://www.cs.waikato.ac.nz/ml/weka/
Course Objectives
Concepts:
1. The student should understand the types of problems being attacked by data mining, particularly through machine learning methods.
2. The student should understand the methods and techniques used to attack data mining problems, particularly machine learning methods.
3. The student should understand how the assumptions made influence the learning methods that can be used.
4. The student should understand in detail various learning methods.
5. The student should understand the methods of evaluating data mining approaches and applications.
Applications:
1. The student should be able to prepare data for data mining – including putting data into a standard format.
2. The student should be able to identify machine learning algorithms that have the potential to address a given data mining task.
3. The student should be able to analyze the results of data mining experiments and come to conclusions
4. The student should be able to write a program recreating an existing data mining method – demonstrating a detailed understanding of the method.
5. The student should be able to integrate concepts from the course to carry out a complete data mining project.
Tentative Course Plan:
|
Date |
Material |
Reading |
Assignments (Tentative) |
|
Aug 30, Sept 1 |
Intro to Class, Intro to Data Mining |
|
|
|
Sept 6,8 |
Intro to Data Mining |
Chapt 1 |
|
|
Sept 13,15 |
Concepts, Instances |
Chapt 2 |
Data Prep Assigned |
|
Sept 20,22 |
Attributes, Data Preparation |
|
|
|
Sept 27,29 |
Output Knowledge Representation |
Chapt 3 |
Mining 1 Assigned |
|
Oct 4,6 |
MIDTERM |
|
|
|
Oct 11,13 |
OneR |
Section 4.1 |
Program Assigned |
|
Oct 18,20 |
Naïve Bayes |
Section 4.2 |
Mining 2 Assigned |
|
Oct 25 |
FALL BREAK – NO CLASS |
|
|
|
Oct 27, Nov 1 |
Decision Trees and Decision Rules |
Section 4.3, 4.4 |
Project Assigned |
|
Nov 3, 8 |
Regression; Instance Based Learning |
Section 4.6; 4.7 |
Evaluation Assigned |
|
Nov 10,15 |
Evaluation |
Chapt 5 |
|
|
Nov 17,22 |
Evaluation |
|
|
|
Nov 24 |
THANKSGIVING – NO CLASS |
|
|
|
Nov 29, Dec 1 |
Engineering input and output |
Chapt 7 |
|
|
Dec 6,8 |
Engineering input and output/ Project Presentations |
|
|
|
Dec 15 10:30 |
Final Exam |
|
|