NAF, PAF - PowerPoint PPT Presentation

1 / 27
About This Presentation
Title:

NAF, PAF

Description:

... by matching engine & corrected ... Non-Matching data: Examples. 05 Mina Rosa St, STRATHFIELD NSW 2135: no match due format ... Non-Matching data: Examples ... – PowerPoint PPT presentation

Number of Views:307
Avg rating:3.0/5.0
Slides: 28
Provided by: aust76
Category:
Tags: naf | paf | nonmatching

less

Transcript and Presenter's Notes

Title: NAF, PAF


1
NAF, PAF AMC
  • Chris Reynolds
  • Manager
  • Network Data Management
  • Australia Post
  • MMUA 25 July 2001
  • Email chris.reynolds_at_auspost.com.au ph
    (03) 9204 7572

2
Objective
  • Outline NAF, PAF AMC (now NDM
  • Explain Quality Levels of data
  • Outline work in hand to improve on Quality
    Address Matching

3
Format
  • What is NDM AMC what does it do?
  • What is the NAF PAF?
  • What is the quality level of the data?
  • Why dont you get a 100 match to DPID?
  • What is Post doing to help?
  • Where can I get more information?

4
Network Data Management
  • NDM is responsible for managing data used in
    sorting mail in the new network
  • 10 M addresses
  • Delivery details for each address
  • Sorting plans for equipment
  • OCR directory retrieval (by liaison with
    suppliers)
  • Support for commercial products (inc PAF)
  • Objective Maximise sorting potential

5
What is the NAF PAF?
  • The NAF is the National Address File used
    internally by Post
  • The PAF (Postal Address File) is the child of
    NAF. It is a subset of the NAF.
  • PAF is extracted on a 3 monthly basis
  • PAF excludes some data contained in NAF
  • Questionable data is filtered
  • Some synonym data excluded

6
Data use
  • Address data is used by both Post and customers
  • Post needs a record of ALL addresses used on mail
    in order to sort mail
  • Customers need a similar record to to barcode
    mail
  • Barcoding is good for Post customers
  • Reduces work for Post
  • Reduces cost for customers

7
Sorting to Delivery Point
  • For Posts own sorting
  • OCR equipment reads addresses on mail items
  • Computers match the address to that in a
    directory file and identify the DPID
  • The DPID bar-code is printed on the mail piece
    and used for sorting the item
  • Bulk customers match addresses and append the
    DPID in a separate bar-code used by Post to sort
    the item

8
Sorting Schematic
DPID Section Delivery Round
Allocation of DPIDs to each machine stacker
9
Quality of NAF PAF
  • Two essential measures
  • Completeness Is every address contained in the
    data?
  • Accuracy Is every address contained in the data
    accurate?
  • Additional for PAF Timeliness
  • How recent is the PAF?

10
Completeness How many of the total addresses
are included, regardless of accuracy?
  • Some address cannot be included due to lack of
    structure. eg Blossom Cottage (no number) or
    123456 Nr STRATHALBYN SA 5255 (Geocode address)
  • Some just missing
  • Some areas better than others
  • Generally better in urban areas than rural
  • Overall estimate is that 2-3 of correctly
    structured addresses are missing Based on
    extensive testing against reference databases

11
Completeness
12
Accuracy
  • Incorrect Locality
  • Locality boundary recently changed
  • Locality boundary unclear
  • Locality incorrect
  • Incorrect road name
  • Road name change
  • Dual road naming
  • Incorrect Spelling
  • Error in data entry, etc.,

13
Overall Quality
14
PAF Timeliness
15
Why dont you get a 100 match?
  • Some blame rests with Post data!
  • BUT. What about your data??
  • Is your address format correct?
  • Is the address complete? (Street nos, etc)
  • Is the address correct?
  • Locality
  • Postcode
  • Spelling
  • Etc

16
Your Address data
17
Mismatch!!
  • Most mismatch errors are in customers data Bad
    structure, Postcode, Locality (vanity
    localities!) and spellings.
  • Some errors are overcome by matching engine
    corrected
  • If Remote or rural addresses predominate then the
    match will be lower, etc..

18
Matching Data!!
19
Database errors
20
Non-Matching data Examples
  • 05 Mina Rosa St, STRATHFIELD NSW 2135 no match
    due format
  • 5 Mina Rosa St, STRATHFIELD NSW 2135 no match
    should be ENFIELD 3136
  • 5 Mina Rosa St, STRATHFIELD NSW 2136 will match
    and correct locality as postcode is correct
  • 5 Minarosa St STRATHFIELD NSW 2136 no match
    ..correct postcode but road is spelt incorrectly
    locality wrong
  • 3 Parry St Wentworthville NSW 2145 No match as
    could be either 3 Perry St Wentworthville 2145 OR
    3 Parry St Girraween NSW 2145

21
Non-Matching data Examples
  • 3 Albert St , CURRIE KING ISLAND 7256 no match
    as Currie is a locality in TAS 7256
  • !23 St Georges Terrace, PERTH WA 6000 no match
    as the address is Saint Georges Terrace! Also
    MT/Mnt
  • RACV, 550 Princes Hwy no match as correct
    address is 542-554!
  • 21st floor Central Plaza One, 345 Queen St
    Likely only to match to primary point despite
    floor 21 being in PAF

Also false matching is possible
Example Actual Betto. In Database Recorded as
Beppo Phone transcription Match
Betts.single character change only!!
22
How do you improve match rates?
  • Analyse reasons for problems
  • Classify prioritise resolution
  • Run a clean up process

23
What is Post doing to help?
  • Quality of Post data
  • Staff input
  • Delivery staff Data use integrated into
    operations
  • Audits regular special checks
  • NDM procedures regular reviews/updates
  • Reference database comparisons
  • Specific Collection projects SchoolsHosp
    currently
  • Data Agreements
  • Land Authorities
  • GNAF

24
What is Post doing to help?
  • Addressing
  • Australian Standards
  • ICSM Street Addressing promotion
  • Rural Addressing
  • postal support
  • Vic completion target mid 2003

25
What is Post doing to help?
  • User Support
  • support thru AMAS licencees
  • PAF users support MMUA site www.majormailusers.
    com.au
  • Specific initiatives
  • New synonyms in PAF
  • assist in address matching
  • New border synonym table
  • for users AMAS licencees
  • Road Alternateswatch this space!

26
Summary
  • Post is committed to improving Quality of the
    NAF/PAF actively working on the task
  • Most matching problems are due to poor client
    address quality
  • Additional processes can improve matching
  • Post is in process of providing additional data
    to assist customers

27
Final Life isnt this bad!!
  • Seek help!
  • Your AMAS supplier
  • MMUA PAF user site
  • Post
  • Customer Connections
  • Network Data Management
Write a Comment
User Comments (0)
About PowerShow.com