Course Software and data
The Nada /misc directory has afs address /afs/nada.kth.se/misc
- XGobi at /misc/tcs/xgobi (installed for Sparc/Solaris 2)
- CoCo : at /misc/tcs/coco/coco.1.3.Beta.r3.Solaris
- S at /misc/tcs/datamining/S
- Biocomputing examples at /misc/tcs/compbio
- Local copy of Irvine ML repository /misc/tcs/mlrepository
- Machine readable data analyzed in Clevelands book /misc/tcs/datamining/cleveland
- Xmdv tool at /misc/tcs/datamining/vizual/Xmdv
Some projects (more to come, I hope)
Clustering image pixels
Papers distributed :
Aug 22: A2,A3,B2-B5(Ch 1-4), C1-C5, D2-6, G1-G2
Sept 12: E3, E2, G3, G4
Sept 19: D7-D9.
Sept 26: F1-F3
Reading until Sept 5:
A2, A3, B2, B3 Fayyad Ch 1,2,(3),4
Reading until Sept 12:
B5, Ch 1-3; C1-C5; D2; E3 . In Fayyad: Ch 6, 7, 11, 12, 13,
Reading until Sept 19:
Reading until Sept 26:
E1-3 (again) ,
Reading until Oct 3:
C3, C5, F1-F3
Find interesting KDD resource pointers and mail to me. If you have
Make your own KDD home page!
Make a proposal for your own examination(evaluation). One or more
Project: discuss project goals and plans with me(SA), staff the
group and prepare a (short or long) class presentation. Projects
after course ends should be presented and communicated to class
by some WWW method.
Paper: Analyze a general problem area, and make your own review
conclusions. Preferably presented in class. Possible topics:
Do Bayesians really own the truth?
True and false knowledge in data mining.
Relations between user and miner
Presentation: Choose an interesting paper or set of papers within
area of KDD-DM. Give a mini-seminar in class, consisting of a summary
by analysis and discussion.
IT IS TIME to start thinking about examination ......
The available slots must fill up before
they occur in real time. The examination is built on the idea
of active and mutual learning.
- Aug 22: Presentation of participants and course content, detailed
- Sept 5: 13:15:The Bayesian method in knowledge discovery and
( Ch 3, 4, B 2-3) (room 1537, PDC)
- Sept 12, 10:15: Presentations of papers on Bayesian approach to
fitting (Sivia Ch 6) , Rough sets and clustering (discussion),
groupwork on examination projects and problems (room 4618, CID)
- Sept 12, 13:15: "Learning to Recognize Volcanoes on
Lars Asker will talk about the development of JARtool,
an image database exploration tool. (room 1537, PDC)
- Sept 19, 13:15: On Bayesian knowledge discovery and stochastic
complexity with an application
to clustering of binary vectors.
Timo Koski, Mathematical
KTH, will talk about experience from a project on classification of
(room 1537, PDC)
- Sept 26, 10:15: Lars Arvestad, Nada:
HMM in biochemistry
11:15:Jakob Eriksson, D2: Grammar Extractor project
Mats Andersson: Method presentation
(room E36 Nada plan3)
- Sept 26, 13:15: Anders Holst, Thesis defence (Bayesian networks
neurocomputing) Kollegiesalen (not formally part of course)
- Oct 3, 10:15-12:00: Daniel Fagerström, CVAP: Time series anaysis: (E51 OB 14)]
- Oct 3, 13:15:Anna Bergman och Ola Ahlqvist: (room 1537, PDC)
Rough sets and attribute oriented induction:
To reduce the number of tuples/rows in a set of task-relevant data stored in
a relation table to mining knowledge rules for characteristic, discriminant,
association, cluster rules etc.
(room 1537, PDC)
- Oct 10, 13:15 Asa Rudström, DSV: (1537, OB 2)
- Dec 5, 13:15:in 1537 (PDC, Osquars Backe 2):
Video show with informal project reportings:
Xgobi: Dynamic Graphics for Data Analysis
Grand Tour and Projection Pursuit
Exploring Time Series Using Interactive Graphics
Spatial CDF Estimation & Visualization with Applications to Forest
Dynamic Graphics in a GIS: Analyzing and Exploring Multivariate Data
Missing Data in Interactive High-Dimensional Visualization