Grid Enabled Microeconometric Data Analysis (GEMEDA)
Project page: GEMEDA
Research on the economic welfare of ethnic groups within the UK is hampered by data availability. Survey data yield minority samples that are too small for meaningful results to be obtained. Census data can provide relatively large samples of minority individuals and households, however, they contain only limited information on economic outcomes, particularly determinants of household income.
In this project a micro-econometric analysis combines both survey and census data (British Household Panel Survey BHPS 1991, UK Census' Samples of Anonymised Records SARs). The analysis is done using Grid enabled data sources and the computational power of Grid HPC resources. For further information please look at our project web page.
The Graphical Side
Part of the project was to develop a graphical user interface for displaying the resulting data. As this project is a web based application, the resulting data are displayed in a Java applet. Java makes it also possible to run under all operating systems. The raw map data was provided by UK Borders. The maps did not quite match the regions needed in the calculations and so a small number of Java programs were developed to combine maps or features from maps. The maps were used in a shape file format (*.shp). A useful library to display and process geographical data files, such as shape files, is the Geotools java library. In this project we used Geotools 2.0.
The Procedure from a Graphical Perspective
Before submitting a job to any of the HPC Grid computers the user has to make a number of choices about the calculations they want to perform. See the following picture for an impression of the input user interface:
After a job is submitted to any of the Grid nodes the user can check the status of their jobs. When a job is finished a result file is returned from the computing node. A program is called to convert between the output format of the calculation process to the required input format for data in the Geotools shape file reader. The so created files get a unique name which is passed as an argument for the applet. Via a link on the web page the user can access this applet. When a applet is loaded both map and data files get access remotely (the stay on the server).
The resulting data can be viewed in two levels of detail, in a regional map and in a SARs map, see the following two pictures:
In the map areas get displayed in white if either no data is calculated for this areas or if the count of people in this area with the selected ethnicity is too low. The user can switch between different ethnic minorities and between sexes; the map will change accordingly. By clicking in a region in the map a box plot will be displayed showing the calculated monthly income for all ethnicities of the selected sex in that area, by clicking outside of a region the values for the whole of the UK will be shown in the box plot. There is a button for displaying the other level (Regional map or SARs map) and a button for viewing a table with the calculated values of the displayed level.
- Ken Clark : Economic specification
- Pascal Ekin : Middleware and databases
- Anja Le Blanc : User interface
- Simon Peters : Computational econometrics
- Stephen Pickles : Infrastructure manager
1. School of Social Sciences (Economics) 2. Manchester Computing, University of Manchester