-
- Geomorphometrics
- Considering Slope, Aspect and how to depict aspect:
- The sine and cosine of the aspect angle give orthoganol reflections of the aspect.
- Maybe it would be good to do the sine of the angle but vary the angle to being from north to being south at regular angles and animate the result.
- My thinking is that this would be like a bright light moving around the entire region.
- Liaising with Robin Smith about a response to a "Call for the Establishment of a European Persistent Geospatial Test-Bed for Research and Teaching"
- Robin recommends that the University of Leeds joins Agile if it is not already a member.
- I am to read the call in detail, have a think and email Robin.
- Browsing
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- MoSeS mapping
- There is a problem with the OA with ZoneCode or ONS_LABEL 00DAFY0013:
- It seems there are two geometries with the same ONS_LABEL, i.e. a split zone. Only one is being coloured.
- Modified mapping code to cope with multiple geometries with the same ONS_LABEL.
- Modified mapping code to link data and plot the linked attributes rather than a random number.
- In a first glance through all the maps it appeared that some OA have high errors in many variables.
- Developed HTML generation code to provide web pages containing images of errors of all Optimisation Constraint Variables.
- Considering how to add Titles and Legends...
-
- Geomorphometrics
- Showed Brian Irvine the Crete images and said I would zip these up and make them available online...
- Error encountered in UK run:
-
Initialised result[14]
Done chunkRow 0
Done chunkRow 1
Done chunkRow 2
Done chunkRow 3
Done chunkRow 4
Done chunkRow 5
Done chunkRow 6
Encountered an OutOfMemoryError initialising Grid2DSquareCellDoubleChunkAbstract 726 out of 4418
java.lang.OutOfMemoryError: Java heap space
java.lang.OutOfMemoryError: Java heap space
- Some method is not coping with OutOfMemoryError correctly
- Set a run going for SRTM90 Mediterranean area.
- Browsing
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- http://edina.ac.uk/projects/seesaw/seegeo/outputsandactivities.html
- Meeting with Paul and Anthony Beck
- Anthony working on VISTA and is employing OGC standards.
- Paul met Anthony by chance on the train and got talking...
- Beck A.R., Fu G., Cohn A.G., Bennet B., Stell J.G. (Forthcoming) A framework for utility integration in the UK
- Recalled the Leeds Humanities Research Institute Meeting 2006-06-07
- Anthony used to work with Duncan who presented that day...
- Anthony suggested I get in contact with Alan Third who is interested in geospatial ontologies and works in Leeds too. I have sent him an email.
- Anthony and Paul both encouraged me to keep blogging the way I am, although Paul wishes I sed something like Wordpress as he has to rely on a search engine for a search facility :-)
- Paul helped find what was wrong with my mapping code causing blank output :)
- _FeatureTypeStyle.setFeatureTypeName( _FeatureTypeName ) was using the wrong _FeatureTypeName:
- Replacing "FeatureTypeName" with "England_oa_2001__area" fixed the problem.
- In fact not setting the _FeatureTypeName also works, so I assume that _FeatureTypeStyle picks up the default as some _Feature.getFeatureType().getTypeName().
- Hurray, I now have colour images!
-
- Browsing
- GRADE
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- Meeting with Paul
- Got a copy of Paul's MappingPortlet.java which has the code I need to generate map images.
- I am very pleased to be working with Paul :)
- Downloaded CS0044 for England Wales and Scotland from CASWEB.
- The table is not available at Output Area level for Northern Ireland.
- Writing loader for CS044
- CAS044DataHandler
- CAS044DataRecord
- There are 110 cells in this cross tabulation.
- It would be a considerable amount of work to write a loader for the entire table.
- Given that only the first column is wanted for now and this is likely to be a one off for the purpose of a test to show that the GA works about as well as IPS for a simple example, then only the first column will be taken care of.
- After a check a simpler table with just the NS-SeC of Household Reference Person (HRP) was not found.
- CASKS014a, CASKS014b, CASKS014c provide for the same variables with more breakdown, but are counts for all people, not just HRPs
- MASS meeting
- Discussing Validation
- People
- Dianna Smith, Karyn Morrissey, Kirk Harland, Andy Evans, Andy Turner
- Validation was thought of as the process of examining and confirming a good fit of variables that have not been constrained/optimised on.
- Not sure this is strictly what I would regard as validation...
- Concerns with over constraining.
- Kirk explained the process of Iterative Proportional Fitting (IPF)
- It basically involves re-weighting
- A set of constraint variables are chosen and ordered.
- Each variable is taken in turn and the ratio of the counts of each value to all values is multiplied by the initial weight for each record where the count is relevant.
- For example if there are 10 people age 20-24 out of a total of 51 then the weights for those in the sample of that age group are multiplied by 10/51.
- This process can be repeated iteratively.
- The 51 records in the sample with the highest weights are those selected.
- Weights in a sample survey are supposed to represent the likelihood of the person/record occuring in the total population.
- I would disregard these weights entirely.
- Instead one could use a random or fixed value.
- May be some variation in results could be obtained by adding randomness to the weights initialised in the survey.
- Is density a useful measure of obesity? Presumably it is and that volume and weight are easily measured although volume is easier done with immersion in liquid.
-
- CSAP meeting
- Andy Turner's CSAP Meeting 2007-05-23 Web Page
- The powers that be decided that from this CSAP meeting onwards no blogging of meeting details is allowed.
- Only official minutes are to be internally disseminated.
- Ho hum! Well at least they final have their act together to have minutes!
- Browsing
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- MoSeS meeting with Mark
- He has done some analysis to compare GA and IPS outputs:
- The GA results are "better" in terms of Car Ownership, but worse in terms of Educational Achievement.
- This is probably down to the differences in the variables used as Optimisation Constraints more than the heuristic.
- Mark is to send me a list of variables so that I can try to Optimise Constrain on the same variables.
- Tables used:
- The variables in these tables have been aggregated and turned into ratios.
- For the purposes of starting to compare the methods I am to optimise on the counts rather than the ratios, but for the same variables.
- NCeSS e-Infrastructure
-
- Browsing
- NCeSS Strategy Board Meeting
-
- UoL
- SoG
- Chat with Sorain Ramchunder who is working with Lee Brown and Joe Holden investigating land management effects on stream ecology.
- How to get detailed spatial information about the areas burned and when in the last 10 years in Moorhouse.
- The benefits of web pages.
- Browsing
- Preparation for my SRDS meeting with Mark:
- Email Belinda, Paul, Steve and Mark for feedback...
- File away my email and go through my Blog for the last year:
- This could take some days...
- Geomorphometrics
- Fixed problems with handling OutOfMemoryErrors
- Results successfully generated for Crete based on SRTM90:
- Results being generated for UK based on SRTM90:
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- Attempting to automate mapping for MoSeS reports:
- Still struggling to get an image on screen or written to a file...
- Preparation for preparing a proposal for the next generation of MoSeS:
- Email exchange with Georgios.
-
- Geomorphometrics
- Fixed issue identified yesterday...
- SoG
- GEOG5060
- Meeting Lindon Tweedle
- Program ran as per instructions with Lindon's account.
- We changed to use a directory on the C:\ recompiled the program and it ran fine.
- We discussed what was needed to improve the groups submission.
- Based on what Lindon says has been done I am sure the group have done enough to pass.
- We identified an issue in that the results seems to be being cropped such that all but a squarish portion of the dataset is returned with a noDataValue. This might be something to do with the input data, but I think there is a bug in my code that should be fixed for next year.
- I am still prepared to review your work before giving a final assessment.
- Browsing
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- Preparation for preparing a proposal for the next generation of MoSeS:
- NCeSS Nodes Commissioning Specification 2007
- Emailed Catriona and Georgios (cc MoSeS core workers in Leeds) to see if they would be interested in developing a proposal together.
- Meeting with Mark
- Mark is considering putting something together with Mike Batty on agents.
-
- Titto Assini and Paul Williamson might also be interested in working together with us.
- I can't find Titto's UKDA Home Page, perhaps he doesn't work there anymore...
- Problem with having too many people involved is that the funding gets so diluted that it is hard to get sufficiently large chunks of work to make all the effort worthwhile.
- Summary of MoSeS work:
- Focus on developing demographic data and models of the UK from 2001 to 2031.
- This microsimulation task has three components:
- Population initialisation at individual and household level for 2001 based on 2001 UK Human Population Census data.
- Addition of non-census variables.
- A dynamic simulation at an annual time step from 2001 through to 2031
- Example applications of the use of the demographic models and data in health service planning are being developed.
- Health service planning can be thought of as providing a Use Case for the development of the demographic models.
- The approach to this work is one of open e-Science:
- To collaborate with experts and build on others work.
- To automate the data processing as much as possible and ensure that our results can be re-generated and something of them are archived for future use.
- At every stage we have tried to expose as much as we can about what we are doing.
-
- SoG
- GEOG5060
- Went to Masters Computing Lab to check everything still works as one group are having trouble running the program:
- All seems fine.
- Meeting Lindon Tweedle at 10:00 tomorrow to run through this together.
- Meeting with Junaid
- We discussed different types of Web Services and their use for Geomorphometrics.
- Browsing
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- GeoLinking IE teleconference
-
- Browsing
- Earth Energy and Environment (EEE) Seminar Transport and Emissions: To Tax or Not
- Peter Mackie
- Introduction
- ~25% CO2 emissions are from transport
- ~90% of this is road traffic
- In 2005 traffic increased 22% and speed decreased 5%.
- Stern report
- Policy questions:
- Do road users pay enough?
- There is an amount of collective payment and individual or classed payment.
- Since 2000 real fuel duty costs have fallen, but overall costs have increased.
- Forge - model used by the government transport department
- Barker report
- Important for land-use planning.
- Discussion
- The best is not the enemy of the good.
- It is simple and easy to add fuel duty compared to road pricing.
- Interesting seminar. I wonder if MoSeS can help by developing an individual level agent based model.
- Microsimulation Seminar Follow Up Meeting
- People
- Martin Clarke, Mark Birkin, Mel, Karyn, Dianna, Me, Paul Norman, Graham Clarke, Andy Evans (arrived late)
- IPF and Simulated Annealing
- Review Paper on Microsimulation
- GC noted that EGC have created a 2 year post-doc position for someone to develop software.
- This is supported with SoG funds.
- HoS AB encouraged CSAP to make a case for a similar position.
- MC to draft a 2 page proposal.
- DS mentioned MPlus
- Alison Hepenstall has written a draft article on MicroSimulation and Agent Based Models.
- Paul Williamsons work on validation
- Ian Shuttleworth dataset for validation
- Economic Activity
- Journey patterns of Protestants and Catholics in Northern Ireland
- Presentation at AAG2007
- Graham and Dimitris Ballas drafted a book chapter for Stewart Fotheringham
- This can be made available.
- Andy Evans asked what is happening with the CSAP proposal for someon to gather all the code for a review:
- Action on Andy to review.
- Dimitris Ballas Geography Matters
- I wondered out loud if we should be thinking about organising hardware upgrade of the Beowulf and/or configuring the Master Computing Lab as a parallel machine.
- Andy Evans reminded us about the MASS meeting next Thursday focussing on validation.
- Geomorphometrics
- Same StackOverFlowError as yesterday... Need to think of a better way of doing this...
- SoG
- GEOG5060
- Chat with Peter Lythe who was looking for Steve Carver to discuss 50% Project
- Couple of ideas:
- Changing gardens and green space in Leeds and its impacts on wildlife and hydrology.
- Configuring a WMS to view Geomorphometric results.
- I showed the Marked work his group submitted and gave some feedback.
- Alison pointed to a Ecological Complexity Special Issue Call for Papers on Environmental micro-simulation: From data approximation to theory assessment.
- David Bell pointed to an Upcoming MISST Event: Everyday Life in the Global City
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- MoSeS meeting with Mark
- Discussed data output formats and transfer.
- We are both of the opinion that the following fields are sufficient:
- AreaCode, PersonID, HouseholdID, HouseholdSARID, IndividualSARID
- I wonder how best to link Jin's data on Age
- SYNPOP
- MoSeS Web Site
- Motivated by upcoming visit by Derek Wanless
- Where to put it?
- What to put on it?
- Looked at my MoSeS Web Page
- Mark thinks the style needs to be improved (he would want more colour).
-
- Browsing
- Distributed Systems/Grid Computing Seminar
- SoG
- Gary Wainman
- Configured and ran a program to harvest Houseprice data for Hull and provided results.
- Geomorphometrics
- Error:
- Initialised result[40]
- Done chunkRow 0
- Done chunkRow 1
- Done chunkRow 2
- Done chunkRow 3
- Encountered an OutOfMemoryError initialising Grid2DSquareCellDoubleChunkAbstract 430 out of 4418
- Done chunkRow 0
- Done chunkRow 1
- Done chunkRow 2
- Done chunkRow 3
- Done chunkRow 4
- Done chunkRow 5
- Done chunkRow 6
- Done chunkRow 7
- Done chunkRow 8
- Encountered an OutOfMemoryError initialising
- Grid2DSquareCellIntChunkAbstract 880 out of 4418
- java.lang.StackOverflowError
- BUILD SUCCESSFUL (total time: 1,176 minutes 24 seconds)
- Not sure how to go about tracking this one down just yet as it is reasonably expensive!
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- MoSeS meeting with Martin
- Recap:
- We started looking at latest results and there was a concern over unemployed males/females age 16 to 24:
- The numbers are quite small, so it may be that normalising by dividing by observed squared would perhaps improve this more than dividing simply by the observed.
- There comes a time when we need to be satisfied with the results enough to move on, now is the time.
- Next step is to focus on mapping out the results and errors to look for anything systematic there.
-
- SoG
- Inter Cluster Event 12:00 to 14:00
- Browsing
- Geomorphometrics
- Ran out of disk space in C:\Temp\output1178907878083 for temporary output files. Either make sure more disk space is available, change the system temporary directory or run with -J-Dnb.output.heap=true
- Changed relevant System Environment Variables TEMP and TMP to D:\Temp instead of C:\Temp and restarted Netbeans.
- Changed Geomorphometrics workspace to use D:\.
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- MoSeS meeting with Belinda
- Belinda asks if I am doing the MoSeS application modelling.
- I'm not really sure what is going on with this.
- Communication
- Belinda finds it hard to make time to develop web pages that detail her work or what she is doing.
- Me too! But I seem to think it is more important and with potentially larger time savings in the long run.
- How to get the data I am producing into Belinda's dynamic model?
- What format?
- Revisiting the problem of coming up with a consistent set of variables from the HSAR and ISAR.
- Looking for data on age specific death rates for Leeds:
-
- Geomorphometrics
- Hit a problem generating data for the UK based on SRTM:
- Done row 5888
- java.lang.StackOverflowError
- at java.util.Vector.ensureCapacityHelper(Vector.java:219)
- at java.util.Vector.addElement(Vector.java:581)
- at java.lang.ClassLoader.addClass(ClassLoader.java:172)
- at sun.misc.Unsafe.defineClass(Native Method)
- at sun.reflect.ClassDefiner.defineClass(ClassDefiner.java:45)
- at sun.reflect.MethodAccessorGenerator$1.run(MethodAccessorGenerator.java:381)
- at java.security.AccessController.doPrivileged(Native Method)
- at sun.reflect.MethodAccessorGenerator.generate(MethodAccessorGenerator.java:377)
- at sun.reflect.MethodAccessorGenerator.generateSerializationConstructor(MethodAccessorGenerator.java:95)
- at sun.reflect.ReflectionFactory.newConstructorForSerialization(ReflectionFactory.java:313)
- at java.io.ObjectStreamClass.getSerializableConstructor(ObjectStreamClass.java:1299)
- at java.io.ObjectStreamClass.access$1500(ObjectStreamClass.java:52)
- at java.io.ObjectStreamClass$2.run(ObjectStreamClass.java:420)
- at java.security.AccessController.doPrivileged(Native Method)
- at java.io.ObjectStreamClass.(ObjectStreamClass.java:400)
- at java.io.ObjectStreamClass.lookup(ObjectStreamClass.java:297)
- at java.io.ObjectStreamClass.initNonProxy(ObjectStreamClass.java:531)
- at java.io.ObjectInputStream.readNonProxyDesc(ObjectInputStream.java:1552)
- at java.io.ObjectInputStream.readClassDesc(ObjectInputStream.java:1466)
- at java.io.ObjectInputStream.readArray(ObjectInputStream.java:1591)
- at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1299)
- at java.io.ObjectInputStream.readArray(ObjectInputStream.java:1634)
- at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1299)
- at java.io.ObjectInputStream.defaultReadFields(ObjectInputStream.java:1908)
- at java.io.ObjectInputStream.readSerialData(ObjectInputStream.java:1832)
- at java.io.ObjectInputStream.readOrdinaryObject(ObjectInputStream.java:1719)
- at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1305)
- at java.io.ObjectInputStream.readObject(ObjectInputStream.java:348)
- at uk.ac.leeds.ccg.andyt.grids.core.Grid2DSquareCellAbstract.loadObject(Grid2DSquareCellAbstract.java:4505)
- at uk.ac.leeds.ccg.andyt.grids.core.Grid2DSquareCellAbstract.loadObject(Grid2DSquareCellAbstract.java:4493)
- at uk.ac.leeds.ccg.andyt.grids.core.Grid2DSquareCellAbstract.loadObject(Grid2DSquareCellAbstract.java:4448)
- at uk.ac.leeds.ccg.andyt.grids.core.Grid2DSquareCellAbstract.loadObject(Grid2DSquareCellAbstract.java:4454)
- at uk.ac.leeds.ccg.andyt.grids.core.Grid2DSquareCellAbstract.loadObject(Grid2DSquareCellAbstract.java:4454)
- ...
- Attempted again having altered Grid2DSquareCellAbstract:
- loadObject(File,boolean) modified so that it swaps all chunks rather than just one when it encounters an OutOfMemoryError.
- This may have implications for the efficiency of other methods, so some optional handling might be a better solution...
- The same problem occurred, so need to debug and try to figure what is going wrong...
- Logic error located, the problem was in trying to handle OutOfMemoryError deep in loadObject(File) this has been fixed, and previous modification changed back.
- Browsing
- DSOnline
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- NCeSS e-Infrastructure Project publication:
- On NCeSS intranet:
- Daw M., Proctor R., Lin Y., Hewitt T., Jie W., Voss A., Baird K., Turner A.G.D., Birkin M., Miller K., Dutton W., Jirotka M., Schroeder R., de la Flor G., Edwards P., Allan R.J., Crouchley R. (2007) Sustaining an e-Infrastructure for Social Science. Abstract submitted to the 3rd International Conference on e-Social Science. PDF, Word Doc.
- Daw M., Proctor R., Lin Y., Hewitt T., Jie W., Voss A., Baird K., Turner A.G.D., Birkin M., Miller K., Dutton W., Jirotka M., Schroeder R., de la Flor G., Edwards P., Allan R.J., Crouchley R. (2007) Sustaining an e-Infrastructure for Social Science. Abstract submitted to the 3rd International Conference on e-Social Science.
- Daw M., Proctor R., Lin Y., Hewitt T., Jie W., Voss A., Baird K., Turner A.G.D., Birkin M., Miller K., Dutton W., Jirotka M., Schroeder R., de la Flor G., Edwards P., Allan R.J., Crouchley R. (2007) Sustaining an e-Infrastructure for Social Science. Abstract submitted to the 3rd International Conference on e-Social Science.
-
- Research Interests
- Road Safety, Road Accident Geography
- Browsing
- MASS meeting
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
-
- Developing an intellectual agenda (for upcoming Inter Cluster Meeting):
- Considering, appreciating, using, evaluating, developing and making available geospatial data, information and technology that changes the way people, government and industry operate. Research in this field is largely led by commercial business and free and open source software development. Some of this work involves individuals and groups operating in applied and academic communities.
- Browsing
- UoL
- SoG
- Staff Meeting
- Read Draft Minutes of Academic Staff Forum from 2007-04-24 or 2007-04-25
- Seemed reasonable although not allowed to make public comment due to confidentiality restriction.
First mention of a research cluster reorganisation post RAE.
-->
- Read Minutes of Research Committee Meeting 2007-04-27
- Mentoring Nanlin
- For the Inter Cluster meeting on 2007-04-14 what does international mean?
- Involving more than one country?
- Of interest internationally?
- Involving people from more than one country?
- Relevant to more than one country?
- How can we best organise for this?
- I mentioned that Andy Evans is trying to orgnaise an event about integrated socio-economic and environmental modelling in the summer and that this theme could be discussed on Monday.
- CSAP Cluster Meeting
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- MoSeS Results
- Beowulf Processing
- Encountered a java.lang.ClassCastException:
- java.lang.ClassCastException: [Ljava.lang.Object;
- at uk.ac.leeds.ccg.andyt.projects.moses.process.GeneticAlgorithm_SWR_HSARHP_ISARCEP.getFitness_NSSE_1(GeneticAlgorithm_SWR_HSARHP_ISARCEP.java:843)
- at uk.ac.leeds.ccg.andyt.projects.moses.process.GeneticAlgorithm_SWR_HSARHP_ISARCEP.getFitness(GeneticAlgorithm_SWR_HSARHP_ISARCEP.java:757)
- at uk.ac.leeds.ccg.andyt.projects.moses.process.GeneticAlgorithm_SWR_HSARHP_ISARCEP._Mutate(GeneticAlgorithm_SWR_HSARHP_ISARCEP.java:679)
- at uk.ac.leeds.ccg.andyt.projects.moses.process.GeneticAlgorithm_SWR_HSARHP_ISARCEP._Breed(GeneticAlgorithm_SWR_HSARHP_ISARCEP.java:422)
- at uk.ac.leeds.ccg.andyt.projects.moses.process.GeneticAlgorithm_SWR_HSARHP_ISARCEP._Optimise(GeneticAlgorithm_SWR_HSARHP_ISARCEP.java:350)
- at uk.ac.leeds.ccg.andyt.projects.moses.process.GeneticAlgorithm_SWR_HSARHP_ISARCEP._Optimise(GeneticAlgorithm_SWR_HSARHP_ISARCEP.java:192)
- at uk.ac.leeds.ccg.andyt.projects.moses.process.GeneticAlgorithm_SWR_HSARHP_ISARCEP._Optimise(GeneticAlgorithm_SWR_HSARHP_ISARCEP.java:127)
- at uk.ac.leeds.ccg.andyt.projects.moses.process.IndividualCensus_SWR_HSARHP_ISARCEP_Beowulf.run_OA(IndividualCensus_SWR_HSARHP_ISARCEP_Beowulf.java:318)
- at uk.ac.leeds.ccg.andyt.projects.moses.process.IndividualCensus_SWR_HSARHP_ISARCEP_Beowulf.run(IndividualCensus_SWR_HSARHP_ISARCEP_Beowulf.java:114)
- at uk.ac.leeds.ccg.andyt.projects.moses.process.IndividualCensus_SWR_HSARHP_ISARCEP_Beowulf.main(IndividualCensus_SWR_HSARHP_ISARCEP_Beowulf.java:71)
- at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
- at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:39)
- at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25)
- at java.lang.reflect.Method.invoke(Method.java:585)
- at runtime.daemon.JarClassLoader.invokeClass(JarClassLoader.java:111)
- at runtime.daemon.ThreadedWrapper$1.run(ThreadedWrapper.java:235)
- at java.lang.Thread.run(Thread.java:595)
- Stopping process <16> on node204.leeds.ac.uk
- Not obvious what is wrong here...
- All other results seem to have returned...
- If not all results are returned then from the missing result onwards, the results will be wrong!
- This is not an explanation as to the outliers produced before as it would be obvious in the Control Constraint Graphs.
- Created new account with http://www.java.net/
- SoG
- Meeting with a friend and former colleague Tom Henderson:
- He is working with software called Wings Platinum available via http://www.avstumpfl.com/:
- It is for presentation of multimedia content.
- Here is where Tom's been hanging out http://www.kuckuck.at/
- Tom might be coming back to work with us in the Graphics Unit.
- Events
- Browsing
- Google appreciation:
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- MoSeS Population Initialisation results
- The results generated by changing the fitness function to a more normalised addition:
- These are very similar to the SSE results:
- Made a minor adjustment to the fitness funciton so that the normalisation is by dividing by the (Observed + 1), the addition of 1 to the top of the fraction was irrelevant.
- By inspecting the logs, it seems that improvement is still being made when optimisation is halted.
- This indicates that it might be sensible to optimise for longer
- Set a new run going optimising for 3000 iterations. Hopefully this will complete by Monday...
Mentoring Nanlin
- Nanlin and the Sustainable Uplands team are of the understanding that the Virtual Knowledge Park that they have been using will no longer be available to them from June.
- The VKP has been a complete and supported solution for the project for managing their content.
- In terms of software, I have suggested Nanlin look at the following:
- In terms of hardware, I suggested liaison with Mike Crabtree and the possible need for project funds to contribute to the maintenance of a server.
Updated various web pages.
OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- MoSeS meeting with Paul
- Chatted about using GeoServer instead of Google Map Creator
- Paul keen to get more publications underway.
- Paul is going to do some XPATH stuff to retrieve the URL of the image and present that on a page for the GLS Client.
- I am working on developing Use Cases by outlining scenarios.
- We are still both annoyed about the data and not being comfortable about using Grid Computing resources in China.
- Meeting with Martin
- We looked at these results:
- Martin still concerned about outliers, but agreed the results were an improvement.
- I pointed out that in the main the fits were better for those variables with higher counts indicating that strategically the Sum of Squared Errors might not be the best thing to use.
- Maybe it is better to use an error measure that normalises based on the observed:
- I suggested adding an arbitrary, but reasonably large number before squaring.
-
- Martin suggested dividing by the Observed Squared
- I am to modify the code and produce some results for Leeds based on a Normalised Sum of Squared Errors
- We are keen to see maps of the errors or something that will help us see if the outliers are the same outliers.
- We are keen to identify outliers based on confidence intervals based on standard deviations or the like.
- Beowulf processing
- Added a Normalised Sum of Squared Errors (NSSE) error function to the code.
- Rather than normalising by dividing by the Observed Squared, this normalises by dividing by the (Observed + 1), the addition of 1 is to prevent division by zero problems.
- Stopped the UK run and started a Leeds run using the Normalised Sum of Squared Errors (NSSE) error function.
- Updated various web pages.
- SoG
- Preparation for my staff review in accordance with the Staff Review and Development Scheme (SRDS) process:
- Mark Birkin, my reviewer provided a folder of information about my role and my last review.
- I am to fill out an SRDS form in preparation for the review and pass this to Mark with whom I will agree a meeting date.
- PhD
- Resent Supervision Summary Sheet Document links to Andy and Mark so that the forms can be signed and handed back to Val.
- Manuel Gloor asked me for the coordinates of the amazon river to plot in Matlab... I started looking http://en.wikipedia.org/wiki/Amazon_River, but after a short chat Manuel decided to look for himself :)
- Began preparation for the Research meeting on 2007-05-14
- Began preparation for the Research meeting on 2007-05-14
- Graham emailed to let us know Ryszard's email adress and that his main interest is economic geography, especially technology, innovation, regional planning and science parks.
- Events
- Browsing
- Subscribed to the http://www.neodc.rl.ac.uk/ NEODC News Channel
- OGC Grid Collision, NCeSS e-Infrastructure, MoSeS
- Emailed Chris Higgins to get him to look at GeoServer Improvement Proposal GSIP 16
- It seems there is a JISC e-Infrastructure Use Cases and Service Usage Models Project, but I can't find a project home page...
- All Beowulf Nodes rebooted this morning, so the UK run has stopped.
- The last area farmed out was 36034 with ZoneCode 00CBFX0026.
- Reconfigured and restarted the run where it left off...
- Thinking about linking BHPS and Census SAR data to produce individual and household level datasets for Census Aggregate Statistic Areas
- There are many ways to do this.
- How about adding one variables at a time:
- I think Kim Proctor suggested this approach in the recent CSAP Seminar:
- What she thought was important was to link the data based only (or in terms of relative importance) on those linking variables that correlate strongly with the variable to add.
- This seems like a good idea.
- In doing this, there is a need to know what variables are common between the data sets:
- Age is likely to be one:
- However, this like many variables is likely to be aggregated in different ways in the datasets, so the linking is not straightforward.
- What is the full list of linking variables?
- The correlation between BHPS variables needs to be analysed:
- Again this can be done in any of a number of ways.
- Some regression type analysis seems sensible based on those variables which are linking variables.
- NCeSS e-Infrastructure
- Emailed feedback to Mike about the draft abstract he proposes to submit to the Third International Conference on e-Social Science
- Browsing
- Expenses
- Sent expenses claim for NCeSS showcase to Kenny who quickly responded to my request :)
- SoG
- GEOG5060
- Meeting with Richard Coppell about 50% Project
- He is keen to develop geomorphometrics of convexity and concavity.
- We had a brainstorm and outlined some ideas:
- Comparisons are done with equidistant opposite values.
- A rotation invariant metric can be reasonably straightforward:
- Calculate the concavity or convexity to add based on the area between the traingle of heights.
- If flat or a planar slope there is clearly neither.
- One can distinguish concavities which are hollow like.
- One can distinguish convexities which are peak like.
- One can distinguish concavities which are becoming more or less concave downslope.
- One can distinguish convexities which are becoming more or less concave downslope.
- More can be done with Axis based metrics
- Recall there are those based on a fixed direction and those which are based on a potentialy different axis for each cell.
- For example, North or slope aspect can be used for the axis.
- For any axis there is along the axis concavity/convexity and orthoganol to the axis concavity/convexity.
- For example, in using slope aspect one can look at profile concavity/convexity and cross profile concavity/convexity.
- A problem as with all axis based metric there is a problem:
- What to do about the measurement if the axis is not in a cardinal directions, i.e. aligning with the axees of the grid itself?
-
- Perhaps the best way is to ask for interpolated point measurements at fixed distance.
- Richard has enough to go on for the time being, but will ask for feedback in about a weeks time.
- My new temporary office colleague:
- MoSeS
- Richard Milton has sent us the Google Map Creator Version 1.1 source code :-)
- I have sent it on to Paul and thanked Andy and Richard at CASA.
- The conditions of us having the code is that we don't let it outside NCeSS.
- We have agreed to that, and will discuss in time if we are allowed to open source our code based on what we learn from the Google Map Creator Version 1.1 source code.
- Chat with Mark
- He has not looked at the latest results I posted yet.
- He is still developing his IPS code and generating results as a benchmark.
- I am to start thinking about how to add variables from the BHPS into an Individual and Household dataset based on the SARs
- I've got a few ideas, but no magic bullet
- Linking
- Which Household in the BHPS matches best each Household from the HSAR?
- Which Individual in the BHPS matches best each Individual from the ISAR?
- I wonder what variable distributions we would get if we substitute the BHPS for the SARs and repeat what I have been doing trying to control constrain on somethings and optimise constrain on similar things.
- References from Martin:
- I am expecting full references, but in the meantime this is the best Marting could do:
- Wilson A.G., Pownall C.E. (1973)
- Searching for this I found:
- Eric J. Millera, John Douglas Hunt, John E. Abraham, Paul A. Salvini Microsimulating urban systems. Computers, Environment and Urban Systems 28 (2005) 9–44.
- M Clarke and Einar Holm (1988 I think) In Geografiska Annaler. Seminal review of microsimulation modelling from 57 to 87!
![[Validate RDF]](http://www.geog.leeds.ac.uk/people/a.turner/images/rdf_w3c_button.gif "W3C RDF Validation")
![[Validate CSS]](http://www.geog.leeds.ac.uk/people/a.turner/images/vcss.gif "W3C CSS Validation")
![[An image of Andy Turner]](http://www.geog.leeds.ac.uk/people/a.turner/a.turner.png)
