Here are some more details about assignment 2.

1) Aims and learning objectives of this assignment

- a self-selected task to extend aspects of the course material 
- involves practical aspects of the machine learning problem, i.e.,
	implementing algorithms and/or experimental evaluation of
	algorithms on data set(s)
- exercise written communication skills in motivating, summarising
	and recording work done on a specified task

2) Due date

As you know, the original due date was to have been the end of Week 14.
Following a vote of the class this was then changed to the end of Week 15,
i.e., midnight of Sunday June 22nd. However, since this is now during the exam
period and I have been getting requests for extensions, I have decided to allow
an extra period of 10 days for submission with no late penalties.  However, you
*must* submit the assignment by midnight of Wednesday July 2nd. After that the
late penalty will the total available marks (20), i.e. no marks can be given
for submission after this time.

Due date: Sun Jun 22 23:59:59
Late penalty after Wed Jul 2 23:59:59 is 20 marks (i.e. no marks can be given
	for submission after this time)

NOTE: assignment 2 is worth up to 20 marks for the course total whereas
	the final exam is worth up to 50 marks. So it is your own interests
	to get the assignment completed as soon as you can. Also, don't spend
	too much time on the assignment, just do as much as you can without
	using up too much of your revision time.


3) Submission

The hand-in for this assignment has two parts: files containing program code to
do something interesting with data set(s) and/or results of running programs on
data set(s), and a report on what you did. So you will be submitting an archive
containing files for the first part and a report document for the second part.

There are two options for the first part, either as a gzipped tar archive
or as a zipped archive. Your file should have a filename stem "files" and
extension either ".tgz" or ".zip".

There are three options for the report, a single document in either
PostScript, PDF or Word format. This file should have a filename stem
"report" and extension either  ".ps", ".pdf" or ".doc".

This gives six acceptable permutations of give submission:

/home/cs9417/bin/classrun -give ass2 files.tgz report.ps
/home/cs9417/bin/classrun -give ass2 files.tgz report.pdf
/home/cs9417/bin/classrun -give ass2 files.tgz report.doc
/home/cs9417/bin/classrun -give ass2 files.zip report.ps
/home/cs9417/bin/classrun -give ass2 files.zip report.pdf
/home/cs9417/bin/classrun -give ass2 files.zip report.doc

NOTE: there is a 200Kb limit on the size of your submission. If this
	is a problem then let me know.


3) Group submission

Please make sure a report from a group with more than one person includes ALL
names and student IDs for the group members.


4) Marking criteria

Total:

20 marks (rounded up *or* down to the nearest integer)

Part 1:

5 marks: solving the basic problem as described in the topic
3 marks: extra features, or 1 person solving most or all of a >1 person problem

Part 2:

5 marks describing the problem and your solution
3 marks good presentation and communication

Achievement:

4 marks: difficulty * 4
e.g.

1/5 * 4 = 0.8
4/5 * 4 = 3.2

etc.


Some general guidance:

For the form of the report see my earlier email referring to message 22 on
Message Board: "Assignment 2 - Sample report outline".

Part 1:

Marks will be gained by:

- evidence of good design by breaking down the problem into sub-components,
	so even if you have bugs it is clear that your approach was a good one
- rigorous collection of results
- use of comments and notes to record decisions taken in the process of
	the work

Marks will be lost by:

- programs failing to compile or run
- missing results files
- no clear information on contents of files submitted
- evidence of plagiarism


Part 2:

Marks will be gained by:

- evidence of thorough testing of an idea
- good presentation and summarisation of key results using tables, graphs, etc.
- simple, clear and direct explanations
- well-formatted, well-organised, spell-checked and grammar-checked documents

Marks will be lost by:

- inappropriate length (2 pages should be enough except for figures, etc. 
	which can go in an appendix)
- digression, rambling or waffling to fill space unnecessarily
- errors or inconsistencies in presentation, such as
	- incorrect description of algorithms or their properties
	- poor algorithm selection for a task
	- errors in evaluation like not using an independent test set to
		assess accuracy if this is required
	- statements conclusions not based either on your experimental
		results or referenced sources
	- incorrect or inappropriate use of statistical tests