School of Computer Science

COMP80131: Scientific methods 1 - Scientific evaluation, experimental design and statistical methods (2011)

Scientific methods 1 - Scientific evaluation, experimental design and statistical methods
Level: 8
Credit rating: 5
Pre-requisites: None
Co-requisites: COMP80122 (Scientific Methods 2) , COMP80142 (Scientific Methods 3)
Lectures: 12 lectures
Lecturers: Barry Cheetham , Goran Nenadic
Timetable
SemesterEventLocationDayTimeGroup
w0 Lecture 2.15 Fri 12:00 - 13:00 -
Sem 1 w7-12 Lecture 2.15 Tue 12:00 - 13:00 -
Sem 1 w7-12 Lecture 2.15 Wed 12:00 - 13:00 -
Assessment Breakdown
Exam: 0%
Coursework: 50%
Lab: 50%

Introduction

This course, referred to as 'Scientific Methods 1', considers the design of experiments and observational techniques for testing ideas that emerge from research in most areas of Computer Science. It is intended for all post-graduate research students.

Aims

The main aim is to address the principles of experimental design and observation that underpin research in most areas of Computer Science. This requires fundamental issues of Scientific Methodology to be raised. The concepts of ‘null hypothesis’, hypothesis testing and the measurement of statistical significance must be addressed with a survey of statistical techniques and tools that are available. Evaluation methods ranging from subjective assessment, evaluations of software and formal statistical approaches will be introduced and illustrated by examples.

Syllabus

The unit will have a series of lectures on experimental design, scientific evaluation and statistical methods, reinforced by illustrations and case-studies based on the experience of researchers in Computer Science and other areas of scientific research. Researchers within the CS School have been to provide examples of how research is evaluated in their particular research areas. Practical scenarios will be described and used to provide opportunities for designing experiments, looking at and evaluating data and developing evaluation strategies for representative problems.

Learning and Teaching Processes


There will be a series of 12 lectures reinforced by discussions on the illustrations, case-studies, assignments and software tools. There will be both individual and group-based practical discussions, including lab sessions for practising statistical software. The use of e-learning is not currently incorporated in this course, but materials for lectures and case-studies will be made electronically available.

Assessment


Assessment will require the students to present a written assignment based on the lecture material and the results of some computer based experimental work and analysis.