Transcript Topic iii supporting papers

Topic iii supporting papers
Dan Hedlin
Statistics Sweden
1
Two topics
• E&I, organisation and processdata
• Unit errors (£ instead of £1000)
2
Unit errors
Where is the similar research?
• Detection and correction of unit errors
• Special case of pattern recognition (data
mining)
• More detailed issue: are you worried about
the ’type I error’? (ie erroneously amending
correct values)
3
Organisation
• Distribute knowledge or methods
throughout the organisation?
• If knowledge, how – equal level for
everyone?
• How to write RPMs – specificity vs
generality
4
Processdata
• How much data do we need to collect?
• Only interesting data?
• Suggestion: some basic data and then only
data with a purpose and an owner
5