CS059 – Data Mining

Fall 2013

 

Home

Material

Slides

Assignments

Administrative

Class Hours: Tuesday 13:00-16:00, Room Ι3.
Instructor: Panayiotis Tsaparas (tsap _at_ cs.uoi.gr), Office Β.3

Grades: The grade for the course will be determined by the assignments and the project.

 

Announcements

·       Tuesday February 13: Projects. The deadline for handing in the project is extended for Sunday Feb 23 night – Monday Feb 24 morning. The schedule for the project examinations will be set on Monday afternoon, for the projects handed in. Here are some instructions on how to write the project report. Read the instructions carefully.The project report counts for a significant percentage of the final project grade.

·       Wednesday January 15: The second part of Assignment 3 is now available on the Assignments page of the course.

·       Wednesday January 8: Clarifications for Question 2 of Assignment 3

o   In the use of the liblinear package you should use the option that returns the probability of each class (-b in the case that you use the command line interface). You should classify a tip to the class with the highest probability if the probability is above the threshold θ. The recall value that you will compute is the fraction of the tips that are classified, while the precision value is the fraction of the classified tips that are classified correctly. You will produce 9 (precision, recall) pairs and you will plot them with recall being on the x axis, and precision on the y axis.

·       Wednesday December 18: The first part of Assignment 3 and the Final Project are now available on the Assignments page of the course.

·       Thursday November 28: Assignment 2, is now available on the Assignments page of the course.

·       Sunday November 14: Clarifications the for Assignment:

o   For the second question, what you are asked to produce is the set of (s,t) bipartite cliques, for input values of s and t. The sentence about the “types” and the “interests” of users serves only as a motivation as to why we would like to find such cliques.

o   For the third question, if you want to use WEKA you would need to find a way to load the data in a sparse form. This is not obvious in WEKA, so you may want to use one of the implementations of FIMI instead.

·        Thursday November 14: Assignment 1, part two is now available on the Assignments page of the course.

·        Thursday October 24: Assignment 1 part one, is now available on the Assignments page of the course.