150 likes | 297 Views
BSPS Annual Conference 2006. School Pupil Forecasting – can GIS improve our methods and understanding? Wendy Pontin Norfolk County Council. BSPS Annual Conference 2006. Background Setting the scene Methodology school catchment forecasts school rolls forecasts
E N D
BSPS Annual Conference2006 School Pupil Forecasting – can GIS improve our methods and understanding? Wendy Pontin Norfolk County Council
BSPS Annual Conference2006 Background Setting the scene Methodology school catchment forecasts school rolls forecasts GIS desk top survey of pupil yield from new housing Issues that have arisen Lessons learnt
Background Residence based school catchment forecasts carried out since the early 1990’s – increasingly using GIS 2005 – Audit commission Report November 2005 – Seminar looking at best practise 2005/06 – Pupil Forecasting Project Pilot study to develop methodology to: forecast the number of 0-19 year olds in Norfolk on individual school rolls and on a resident school catchment basis for each year group over a six year forecast period
Setting the scene Over 400 schools: Highs – 52 Primaries – 200 Infant / First – 112 Junior / middle – 70 School size 42 schools of under 50 pupils 160 schools of under 100 pupils High schools vary from 600-1,500 pupils Numbers 105,000 pupils Year R to Year 11 approx. 8,000 in any year cohort
Methodology Brief non-technical description of the actual school forecasting methodology Concentrate on the extensive use of GIS No maps!!! The power of GIS to process and analyse large data sets – a means to an end!
School catchment forecaststhe base data Base year for forecasts Sept 06 Individual pupil records Year R to Year 11 – with valid postcodes Individual pupil records of pupils living in Norfolk attending schools in Suffolk, Cambridgeshire and Lincolnshire – with postcodes FHSA patient register data giving aggregated data to unit postcode level of 0-4 year olds (as of 31 Aug 06) New house build data aggregated to smallest building block level (geo-referenced) – permissions, allocations etc Historic completions data aggregated to smallest building block level (geo-referenced)
School catchment forecastsassemblingthe base data All 2006 base data geo-referenced and taken into GIS Aggregated to: County Children’s Services Areas (5) Current High school catchment areas Current Primary school catchment areas (includes infant, first, junior, middle) Records of 0-4s and school pupil records from Sept 03, Sept 04 and Sept 05 re-aggregated to current school catchment areas etc – already geo-referenced in GIS
School catchment forecastsinto production! Top down approach Using cohort survival roll forward and taking account of children generated by new house build Forecast for County Children living outside the County attending Norfolk schools Children’s Services areas controlled to County High Schools controlled to Children’s Services Areas Primary, Infant and First controlled to High schools Produce by year group (0-15) forecasts for every school catchment area Select appropriate year groups to produce school catchment forecast for each school –robust forecasts!
School roll forecastsdata preparation(1) Sept 2006 Patterns of ‘parental choice’ – or is it??? Produce for each Year group a matrix of ‘parental preference An example (Year 1 say)
School roll forecasts1st attempt Apply Sept 06 pattern of ‘parental preference’ to forecast years for each school Roll forward pattern of parental preference (e.g. for Sept 07 forecast for Year2 roll forward pattern of parental preference of Year1 - Sept 06 to Year2 – Sept 07 and apply to residence based forecasts for Year2 – Sept 07) For entrance year group in any school assume the Sept 06 pattern of parental preference throughout the projection period. This assures consistency with residence based forecasts - does it reflect reality? What next?
School roll forecastsMore data preparation! Look at past admission capacities for each school to actual numbers on school roll in Sept 2006 to set capacity limits for all year groups in Sept 06 Project forward admission capacities for the six year forecast period Assess historically which schools remain at full capacity (oversubscribed) – assume that they will remain so for the duration of the forecasting period For these schools, pupils who cannot get into one of them as 1st choice build up patterns of schools to which they do attend (e.g. of pupils who cannot get into school A, 20% go to school B, 30% to school B and 50% to school C)
School roll forecastsFinalised results For schools that are assumed to be oversubscribed (i.e. up to full capacity) for the duration of the forecast period, adjust results to either bring them up or down to capacity, taking away from or adding the residual from other schools according to the pre-determined patterns – manual adjustment to oversubscribed schools but automatic redistribution For all other schools where overcapacity is identified in any year group, redistribute this to likely neighbouring school – manual adjustment and redistribution Model will automatically adjust succeeding forecast years after any manual adjustment – assumes that pupils stay in a school once they have been allocated there’! Again consistency with residence based forecasts is maintained.
GIS desk top survey of school pupil yield from new housing From districts Housing Land Availability studies (April 2005) digitise boundaries of all recent permissions with 5 or more completions Analyse geo-referenced records of Sept 05 school pupils and younger children aged 0-4 who are resident within these boundaries Analyse by Age Area Type of housing (for the future) Size of permission (i.e. total number of houses) Determine pupil ratios which can be used in the residence based forecasts.
Issues that have arisen Fundamental – accuracy of base data. This means awareness of those administering this data of the importance of accuracy and full coverage Good intelligence of reorganisation (present and future) and of quality of data Using GIS opens up potential and possibilities but still resource hungry (both in terms of computing capacity and people) – processing and analysing very large data sets Need to get buy in from customers – managing expectations! Validation – key to success (resource hungry!!!)
Lessons learnt Methodology proved – can be implemented Methodology proved – seems to be delivering accurate results (pilot results accuracy for first forecast year) Western Children’s Service Area within 1% High schools majority within 1% For the first time we have produced consistent forecasts (resident and school rolls) fit for purpose for a variety of service provision The real test is still to come - a full run for the whole County to very tight time scales Wish us luck!!!