Back to my home page; to my teaching page.

Data Mining and Machine Learning

General Notes for 2012 session

Here is the timetable of lectures and coursework information for 2012

Courseworks A, B, [must do and pass, but no marks] -- Courseworks 1 (30%), 2 (40%), 3 (30%)

Your lecturer is:

Professor David Corne, http://www.macs.hw.ac.uk/~dwcorne/

I want to arrange my office hour in a way that suits the majority of students I am teaching this semester, so please make your preferences known at this doodle poll http://doodle.com/pe9p9e2f2d8bip7u

week	date	Thursday 13:15 EM183	Friday 12:15 WP108	Coursework handout	Coursework handin
1	w/b Mon 10th Sep	DMML intro	basics 1	C/W A C/W B
2	w/b Mon 17th Sep	basics & CW1	basics 3	C/W 1
3	w/b Mon 24th Sep		The a priori algorithm
4	w/b Mon 1st Oct	correlation / regression	Feature selection	C/W 2	C/W A Friday 5^th 23:59pm
5	w/b Mon 8th Oct	Clustering			C/W B Friday 12^th 23:59pm
6	w/b Mon 15th Oct	Naïve Bayes			C/W 1 Friday 19^th 23:59pm
7	w/b Mon 22nd Oct	Decision Trees		C/W 3
8	w/b Mon 29th Oct	Top10DM / Overfitting / Neural Nets / SVMs			C/W 2 Sunday 4^th November 23:59pm
9	w/b Mon 5th Nov	Classifier Ensembles
10	w/b Mon 12thNov	Accuracy, Rules, Text
11	w/b Mon 19th Nov				C/W 3
12	w/b Mon 26th Nov

The course is examined 100% by coursework. Marked coursework includes 3 pieces all set by DWC, .
but two other small pieces of coursework have to be handed in too: A and B, both set by DWC.

OLD THINGS

I am redoing the 2012 DM lectures and courseworks a bit; below is what it was in 2011, and gradually the correct things will be listed in the table above as soon as they are done. In the meantime you can refer to below to get some idea of what is coming up.

thing	slides	coursework handouts	otherthings
DWC Lecture 1	Overview of DM -- Types of data -- examples;	COURSEWORK A (in the slides)	a paper describing some retail basket data
DWC Lecture 2	Some basic issues: classification, 1-NN, normalisation, discretization;	COURSEWORK B (in the slides)
DWC Lecture 3	Basic Statistics and Coursework 1	COURSEWORK 1	Some awk scripts: znorm.awk, cs2ss.awk, fiddlefield.awk, removefields.awk, fixcomdata.awk.
DWC Lecture 4	Basket Data/Association Rules (A Priori algorithm)		the A priori paper
DWC Lecture 5	Correlation and Coursework 3	COURSEWORK 3	The naive bayes awk program for use in CW2 A paper concerned with feature selection when you want to find a very small number of features. Recommended read.
DWC Lecture 6	Cluster Analysis and Clustering
DM Lecture 7	Feature selection		the original relief paper; a paper with many variants of the relief method; the Dash and Liu survey on FS A couple of my papers with PhD students concerning feature selection: -about Text features -about Genes
DM Lecture 8	Similarity and Distance
DM Lecture 9	TBA

Assessment

The assessment of this module is entirely by coursework. My aim in setting the coursework is to get you to learn about the essentials of data mining by working on real data and real issues that myself and/or my colleagues are currently working on. You will implement and use techniques that are in fact quite straightforward, but are nevertheless very important, and in fact these techniques are often misused or not used at all (but of course they should be) by data analysts in science and industry.

Coursework Assignment 1: Worth 40% -- assessed by a report, and in some cases a brief viva.
Coursework Assignment 2: Worth 20% -- set by PC, he will advise.
Coursework Assignment 3: Worth 40% -- assessed by a report, and in some cases a brief viva
Coursework Assignments A, B: Worth 0% -- but you fail the module if you don't submit an adequate response to each. For these courseworks, if I am not happy with what you submit, I will ask you to improve it and resubmit it. This will iterate if necessary until I think you have done it properly.