Interfacing multiple levels of decision support with MapServer to evaluate education quality in Peru Open Source GIS Conference 2004 & Second Annual MapServer User Meeting G. Brent Hall and Michael Leahy University of Waterloo June 9-11, 2004
Jan 09, 2016
Interfacing multiple levels of decision support with MapServer to evaluate education quality in Peru
Open Source GIS Conference 2004 & Second Annual MapServer User Meeting
G. Brent Hall and Michael LeahyUniversity of Waterloo
June 9-11, 2004
Outline
Objectives of project Hierarchical nature of Peruvian education system Design of interface and tools for evaluation of
education quality from multiple perspectives and multiple levels
Role of Open Source Software (OSS) and MapServer in design
Sample decision problems and tools Need for additional spatial analysis techniques in
addition to map-based selection and browsing
Project Objectives
To utilize OSS tools fully without any commercial software other than the OS (as reviewed in Leahy/Hall presentation)
To facilitate flexible evaluation and decision analysis for knowledgeable and less knowledgeable users Knowledgeable users > Ministry planners at 3 levels of
system (national, regional, local) Less Knowledgeable users > School administrators,
teachers, parents To democratize information access and to return
information to collection points (schools)
Project Objectives
To interface map-based selections and list-based selections and to visualise analysis results on maps
To facilitate use of results for planning and system adjustment aimed at improving aspects and areas of education quality deficiency
Peruvian Education System
Ministry of Education
Educational Sub-Regions
Local Education Management Units (UGEL)
Educational DevelopmentAreas (EDA)
Areas of Execution (AE)
Total Non-FormalPrograms
Education Centres(Schools)
Private EducationCentres
Non-Formal PrivateEducation Programs
Non-Formal PublicEducation Programs
Drill down within levels
Drill across between schools within and across levels
Public EducationCentres
Regional EducationDirectorates (DRE)Regional EducationDirectorates (DRE)
National
Regional
Local
Local
Selection
Selection of areas and schools for quality analysis must be flexible and intuitive Should reflect the general organizational hierarchy
of the education system Must eventually allow user to select multiple
areas, single areas, all or only some schools Interface must facilitate users with a range of
computing skills and languages Work remains to be done with this in the prototype
cont….
Currently administrative areas of Ministry of Education are not in digital form (DRE, UGEL) Hence, geographically nested political areas are
used (Departments, Provinces, Districts) Selected entities do not have to be spatially contiguous
to allow user to select them in either map or list mode Currently analysis is restricted to all schools in
selected districts Addition of selection of specific schools to be
developed in the prototype
Database and decision levels
Table of all districtsin Peru
1
n
Selected districts foreducation quality analysis
Selected districts and all constituent schools form current selection set – number of schools is equal to sum of all schools in selected districts
Available indicators at the school level – areaggregated as required depending on level of analysis
1 n
In a typical case a user would select a specific district, a set of contiguous districts, or a set of non-contiguous districts. These selections are saved as a selection setalong with all schools in the relevant areas
Ideally the system should allow selection of one or moredistricts and then allow the user to refine the selection byselecting some or all schools that fall within this selectionset.
These schools can then define the aggregation of urbanblocks or rural population centres that fall within the school catchments to be selected for contextual analysis
cont….
Once a district selection set is defined by a user this currently includes all schools and all enclosed small area census units If the user chooses to work at the district level for
analysis – all census data and all schools data are aggregated to that level – i.e. results become generalized to that level of geography and are not very informative
If the user chooses to work at the school and at the city block or population centre level all data are analyzed at that level, retaining local variability
Database and decision levels
Table of all schoolsin selected districts –all are selected by default
1
n
Available indicators education quality indicators – areaggregated as required depending on level of analysis
1 n
Next step is for user toselect education andcontextual variables
Database and decision levels
Data matrix
Table of all schoolsin selected districts – atpresent all are selected by default
Contextual variables areaggregated either at districtlevel or disaggregated tocity block (urban) or population centres (rural)
1
n
Available indicators at the school level – areaggregated as required depending on level of analysis
1 8
Selection by map
As noted, Ministry administrative unit boundaries are not digitized – political boundaries are used. Areas are spatially nested and analysis canbe undertaken for all schools within selected areas
Selection by map
This map of districtshas been zoomed tothe general level ofthe cono norte of Lima
Districts can beselected in a varietyof ways
Selection by map
Using the select byline tool all districtsin the cono norte have been selected for analysis
Other functions of the EduCal tool
Export results tables to .csv and .shp for further analysis in general and specialist software SPSS, Excel Geographically weighted regression, spatial autocorrelaton
analysis Create and print map composition of results Email maps as attachments (for transmission within
the education system administration) for reporting and planning purposes
Creation, retrieval and modification of user defined scenarios
Planned functions for the EduCal tool
Four areas of planned expansion Modify selection capabilities so user can select
specific schools (single school or groups) for analysis From selected schools define catchments based on
enrolment constrained allocation of city blocks or population centres
Use the census-based entities within these catchment areas for analysis of contextual effects on school selection set
Allows local effects to be analyzed more realistically and flexibly
cont….
Expand data administration functions to allow upload of data from schools to database Requires data cleansing and metadata before data are
added to the database – some automation is possible Needs review of site security and access so authorized
users can control access to their own data (private vs public views of components of the database)
Access should include ability to upload and ability to view via secure password – beyond level of password control that currently exists
Eventual control of site to be passed to counterpart agencies in Peru (Ministry and NGO partners)
cont….
Add dimension of time to create a data cube so quality performance can be analyzed temporally Allow users to perform spatial and temporal queries of
data across and through multiple slices of the data cube Education data are collected from all state schools every
year in Peru hence database design must be extensible Initiate basic data mining functions on-line to allow
planners to return simple queries and examine trends in education quality at the school level prior to and after intervention
Similar work underway in health care sector with JMap (Kheops Technologies) commercial java-based web map server
cont….
Expand spatial analysis functions and mapping output of the tool to enable on-line spatial data analysis (SOLAP) Use results of spatial analysis to inform decision making
process Need to control access to more sophisticated functions for
advanced users only
Access to development site: Access to the developershttp://gaia.uwaterloo.ca/educal [email protected] to public site:
[email protected]://toa.uwaterloo.ca/educal [email protected]