190 likes | 420 Views
Data delivery Eileen Howes 10 April 2006. Summary. What we wanted What we got What we want from 2011 Census. What we wanted. Test data Early release of data for pre-processing To help with QA as data experts CSV files Data that was correct first time Table layouts as agreed in advance.
E N D
Summary • What we wanted • What we got • What we want from 2011 Census
What we wanted • Test data • Early release of data for pre-processing • To help with QA as data experts • CSV files • Data that was correct first time • Table layouts as agreed in advance
Test data • Got some eventually • Had to fight for it
Early release for pre-processing • No • Got the Supertable data on publication date • Had to wait a bit longer for CSV files • Thousands of users had to wait a lot longer
User help with QA • No • So found the errors as soon as we loaded the data
CSV files • Some Supertable • Some CSV files • Some Excel spreadsheets
Data that was right first time • Not always – but a lot to ask • Expert users would have found some of the errors but not all
Table layouts as agreed in advance • No • Some were as published • Others not known until the data arrived • So even more delays in processing • Some were different for different areas
A few points… • Almost all the right numbers • Not necessarily in the right order • Some re-releases of data – extra work but generally OK • Re-releases of just the cells that were wrong –please, NO
A few more points… • Commissioned tables • Some files with no area codes • Because we didn’t specify them • But we expected them to be there
Nightmare at City Hall • Commissioned table C0310 in Supertable • Output area of workplace to ward of residence • Part of SWS Table 301, means of travel to work • ONS offered to split large file for us
Nightmare at City Hall • To export csv files from Supertable: • Tried to export csv but computer crashed • So ran it until it crashed, retrieved it it parts • Started next run from where it crashed
Nightmare at City Hall • CSV file with 15,200,000 records • Then had to add area codes • Area names already there
And another thing… • Area names were not always there • But if the names were not there • Then the codes were • Extra work adding them back in • We need both
What we want from 2011 Census • Standard datasets • Nationally comparable data • Test data • Early release of data for pre-processing • To help with QA as data experts • CSV files • Data that is correct first time • Table layouts as agreed in advance
What we want from 2011 Census • ONS to ditch Supertable • Use something that produces usable csv files • Is easy to use • Is easy to print from
What we want from 2011 Census • Basic standards ALWAYS adhered to • Unique area codes always included • Area names always included • All numbers in the same order for different areas • Agreed table layouts not changed at the last minute
What we want from 2011 Census • Better method of disclosure control • Rounding only if pre-tabulation • Consistent database • Data for administrative areas – wards and parishes • etc