CSC 240/440, Spring 2017, Data Mining
Lecturer: Ji Liu, Office hour: Wed. 4-6pm
TA office hour:
Haichuan Yang (Mon. 2-4pm, CSB 725, firstname.lastname@example.org)
Chuyang Ke (Tue. 3-4pm, CSB 725, email@example.com)
Chester Holtz (Thu. 10-11am, CSB 603, firstname.lastname@example.org)
This courses will introduce fundamental concepts and techniques of data mining, including data attributes, data visualization, data preprocessing, classification methods, cluster analysis, and mining frequent patterns, association and correlation. Advanced topics could be included if the time allows, such as sparse learning, semi-supervised learning, and outlier detection. Graduate students (CSC440) may receive additional questions or requirements in homework or projects.
- Section 0: syllabus [pdf]
- Section 1: Introduction [pdf]
- Section 2: Data [pdf]
- Section 3: Data Exploration [pdf]
- Section 4: Classification: Basic Concepts, Decision Trees, and Model Evaluation [pdf] [A video about Entropy] [probability slides]
- Section 5: Classification: Alternative Techniques [pdf] [linear regression and SVM notes]
- Section 6: Cluster Analysis: Basic Concepts and Algorithms [pdf] [some other materials]
- Section 7: PCA [note]
- Section 8: Association Analysis [note]
- Section 9: Anomaly/outlier detection and missing value imputation [note] [missing value note]
- Section 10: Ensemble methods: XGBOOST [slides] [a good summary]
- Section 11: Semi-supervised learning [slides]
- Homework 1 (Due on Feb. 14) [pdf] [data set]
- Homework 2 (Due on Mar. 2) [pdf] [data set]
- Homework 3 (Due on Mar. 21) [pdf] [data set 1] [data set 2]
- Homework 4 (Due on Apr. 18) [pdf] [data set]
- Homework 5 (Due on May. 9) [pdf]
- Time: Mar. 30 (Thursday) in class. 75 minutes. You can bring the calculator and one cheating paper (letter paper size) with handwriting on both sides.
- It is available on now. [pdf]
- The project presentation is scheduled on Apr. 27 and May 2. The weight of presentation is 20%. The number of total points is 100. Each team will present its project to the rest students in the class, who will grade the presentation performance. Each presentation includes 7 mins for speakers + 3 mins for Q&A.
- The weight of the final project report is roughly 80%. It is due on May 15 midnight, which means that you can update your submission on the leader board until such date. Please submit only one report for each team through blackboard. Your report should include but not limit to 1) the approaches you used for preprocess the data set; 2) all approaches you have tried and their performance; 3) comment why it works or not; 4) what is the best performance you get and the rank of your team on leader board (showing a snapshot is enough)?; and 5) others you think it is novel and important. The weight of the performance of your result is worth 50% points of the report. We use the results on the leader board at 11:59pm on May 15 for the final evaluation. Top three teams in this class will receive 10, 8, and 5 bonus points respectively. If your rank on the leader board is among top 30 or 50, you will receive additional 20 points or 10 points. Note that if your code is found to be copied from anywhere or others (or there exist other cheating behaviors in your submission, e.g., your code does not match your submitted result or some other team uses your code for submission), a heavy penalty will be applied (0 point for the project and it will be reported to the university)!
This course largely uses the course materials from Pang-Ning Tan, Michael Steinbach, and Vipin Kumar. I very appreciate their invaluable contribution and efforts to the data mining field.