# A clear statement of the research objectives of the project.
To understand what can be determined from analysis of real-time train delay data concerning the correlation of delays in different parts of the network.

# A discussion of why it is interesting.
In the railway, delays cause more delays, essentially because we have a system of queues linked by a graph.   Real-time data is now available (but will need to be collected for this project).   I have already developed good models of the distribution of delay, but we do not understand the correlations (temporal or spatial) between delays at all.   Any progress here would be fundamental to further useful output, like better trip-planning algorithms, and improving the network topology.   

# A brief summary of the background to be assimilated and techniques required.
Only simple ideas about networks of queues are needed.   Trains are non-Poissonian, so not much theory is available.   Some skill in python programming is needed to get the data from the web (I have this bit already) and massage it into appropriate forms (to be done).  After that, simple statistical techniques will suffice.

# A list of prospective deliverables.
Some summary statistics on correlations, and software for collecting further data.

# An indication of the relation to end/downstream users:  who should benefit from this [line of] research?
I am in contact with the Railway Engineering Dept in Birmingham University, who are very interested on this type of work.   Eventually we hope to interest Thales, who have the contract to run the National Rail website.

# A brief outline of avenues for a follow-up PhD project.
Possibly more theory on networks of queues with non-Poisson input processes.