Data Points

As a PM for world postcode boundaries I'm often asked do we have boundaries for all the postcodes in a given country?  Or why don't we have boundaries for 6 digit postal codes in Canada or 8 digit postal codes in Brazil?

It's important to understand a bit more about how postal code boundaries are commonly generated.  Remember first that a postal code boundary does not typically exist in reality.  Postal codes are a means of delivering mail and don't need to conform to administrative or natural boundaries.  They are created using a combination of postal delivery points and routes and building blocks of geography.  When you have enough delivery points a boundary becomes possible.   

When you look at the whole of postal codes and types of delivery in a postal system however, you encounter postal delivery types that are unique businesses, post office boxes, military collection centers.   When we create postal products we commonly represent these as points, not polygons because they have specific locations.   So within our 5 digit ZIP code product in order to represent a full roster of postal codes you would be working with both points and polygons.  

For postal codes like 6 digit codes in Canada (local delivery units) and ZIP +4's in the USA, these often represent street sides and multi unit dwellings. Centroids are then generated for a representative location vs attempting to render a polygon for such a localized area.

Finally, our products often include centroids pre-calculated for boundaries allowing you to have a full roster of postal codes as points.   Polygons however are typically reserved for postal codes where they can derived with the right criteria being met.

------------------------------
John Tagliaferro
Product Manager - LI Data Product Team
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

This is very good information to share with Partner and customers.

As a partner manager you are often asked differences between PB data and other vendor  data products. Often to replace existing data at a client.

Recently I had a request for 6 digit postcodes in China and we only had a 4 digit.
By testing our reverse GeoCode API, I could see that we do use the 6 digit postcode, but by you explanaition it makes perfectly sense, that the information is stored on the address point and not as a polygon.
But do we consider to add the 6 digit centroids to countries like China ?

------------------------------
Sissel Kegel
Channel Manager Denmark/Benelux/Baltic
Pitney Bowes Software Europe Ltd.
Denmark
------------------------------

Original Message:
Sent: 03-14-2019 12:25
From: John Tagliaferro
Subject: Post Code Boundaries vs Points


As a PM for world postcode boundaries I'm often asked do we have boundaries for all the postcodes in a given country?  Or why don't we have boundaries for 6 digit postal codes in Canada or 8 digit postal codes in Brazil?

It's important to understand a bit more about how postal code boundaries are commonly generated.  Remember first that a postal code boundary does not typically exist in reality.  Postal codes are a means of delivering mail and don't need to conform to administrative or natural boundaries.  They are created using a combination of postal delivery points and routes and building blocks of geography.  When you have enough delivery points a boundary becomes possible.

When you look at the whole of postal codes and types of delivery in a postal system however, you encounter postal delivery types that are unique businesses, post office boxes, military collection centers.   When we create postal products we commonly represent these as points, not polygons because they have specific locations.   So within our 5 digit ZIP code product in order to represent a full roster of postal codes you would be working with both points and polygons.

For postal codes like 6 digit codes in Canada (local delivery units) and ZIP +4's in the USA, these often represent street sides and multi unit dwellings. Centroids are then generated for a representative location vs attempting to render a polygon for such a localized area.

Finally, our products often include centroids pre-calculated for boundaries allowing you to have a full roster of postal codes as points.   Polygons however are typically reserved for postal codes where they can derived with the right criteria being met.

------------------------------
John Tagliaferro
Product Manager - LI Data Product Team
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

Thanks Sissel,

I'm reviewing options with one of our regional vendors.  They have indicated they do not have the reference data to produce 6 digit postcode boundaries for China but I am now checking if they have points.

Regards,
John

------------------------------
John Tagliaferro
Product Manager LI Data Products
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

Original Message:
Sent: 03-15-2019 04:51
From: Sissel Kegel
Subject: Post Code Boundaries vs Points

This is very good information to share with Partner and customers.

As a partner manager you are often asked differences between PB data and other vendor  data products. Often to replace existing data at a client.

Recently I had a request for 6 digit postcodes in China and we only had a 4 digit.
By testing our reverse GeoCode API, I could see that we do use the 6 digit postcode, but by you explanaition it makes perfectly sense, that the information is stored on the address point and not as a polygon.
But do we consider to add the 6 digit centroids to countries like China ?

------------------------------
Sissel Kegel
Channel Manager Denmark/Benelux/Baltic
Pitney Bowes Software Europe Ltd.
Denmark

Original Message:
Sent: 03-14-2019 12:25
From: John Tagliaferro
Subject: Post Code Boundaries vs Points


As a PM for world postcode boundaries I'm often asked do we have boundaries for all the postcodes in a given country?  Or why don't we have boundaries for 6 digit postal codes in Canada or 8 digit postal codes in Brazil?

It's important to understand a bit more about how postal code boundaries are commonly generated.  Remember first that a postal code boundary does not typically exist in reality.  Postal codes are a means of delivering mail and don't need to conform to administrative or natural boundaries.  They are created using a combination of postal delivery points and routes and building blocks of geography.  When you have enough delivery points a boundary becomes possible.

When you look at the whole of postal codes and types of delivery in a postal system however, you encounter postal delivery types that are unique businesses, post office boxes, military collection centers.   When we create postal products we commonly represent these as points, not polygons because they have specific locations.   So within our 5 digit ZIP code product in order to represent a full roster of postal codes you would be working with both points and polygons.

For postal codes like 6 digit codes in Canada (local delivery units) and ZIP +4's in the USA, these often represent street sides and multi unit dwellings. Centroids are then generated for a representative location vs attempting to render a polygon for such a localized area.

Finally, our products often include centroids pre-calculated for boundaries allowing you to have a full roster of postal codes as points.   Polygons however are typically reserved for postal codes where they can derived with the right criteria being met.

------------------------------
John Tagliaferro
Product Manager - LI Data Product Team
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

This post was removed

To address Sissel's question, we do offer 6 digit postal codes in China, as points with over 40,000 points.  They are part of our Postcodes, Suburb and Admin Boundaries product for China.  It is an annual release, we have 2018.09 available on the data marketplace!

Regards,
John

------------------------------
John Tagliaferro
Product Manager LI Data Products
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

Original Message:
Sent: 03-18-2019 08:47
From: John Tagliaferro
Subject: Post Code Boundaries vs Points

Thanks Sissel,

I'm reviewing options with one of our regional vendors.  They have indicated they do not have the reference data to produce 6 digit postcode boundaries for China but I am now checking if they have points.

Regards,
John

------------------------------
John Tagliaferro
Product Manager LI Data Products
PITNEY BOWES SOFTWARE, INC
White River Junction, VT

Original Message:
Sent: 03-15-2019 04:51
From: Sissel Kegel
Subject: Post Code Boundaries vs Points

This is very good information to share with Partner and customers.

As a partner manager you are often asked differences between PB data and other vendor  data products. Often to replace existing data at a client.

Recently I had a request for 6 digit postcodes in China and we only had a 4 digit.
By testing our reverse GeoCode API, I could see that we do use the 6 digit postcode, but by you explanaition it makes perfectly sense, that the information is stored on the address point and not as a polygon.
But do we consider to add the 6 digit centroids to countries like China ?

------------------------------
Sissel Kegel
Channel Manager Denmark/Benelux/Baltic
Pitney Bowes Software Europe Ltd.
Denmark

Original Message:
Sent: 03-14-2019 12:25
From: John Tagliaferro
Subject: Post Code Boundaries vs Points


As a PM for world postcode boundaries I'm often asked do we have boundaries for all the postcodes in a given country?  Or why don't we have boundaries for 6 digit postal codes in Canada or 8 digit postal codes in Brazil?

It's important to understand a bit more about how postal code boundaries are commonly generated.  Remember first that a postal code boundary does not typically exist in reality.  Postal codes are a means of delivering mail and don't need to conform to administrative or natural boundaries.  They are created using a combination of postal delivery points and routes and building blocks of geography.  When you have enough delivery points a boundary becomes possible.

When you look at the whole of postal codes and types of delivery in a postal system however, you encounter postal delivery types that are unique businesses, post office boxes, military collection centers.   When we create postal products we commonly represent these as points, not polygons because they have specific locations.   So within our 5 digit ZIP code product in order to represent a full roster of postal codes you would be working with both points and polygons.

For postal codes like 6 digit codes in Canada (local delivery units) and ZIP +4's in the USA, these often represent street sides and multi unit dwellings. Centroids are then generated for a representative location vs attempting to render a polygon for such a localized area.

Finally, our products often include centroids pre-calculated for boundaries allowing you to have a full roster of postal codes as points.   Polygons however are typically reserved for postal codes where they can derived with the right criteria being met.

------------------------------
John Tagliaferro
Product Manager - LI Data Product Team
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

John, thanks for the information on how PB represents postal boundaries and postal points.  I know one of the challenges in creating boundaries is deciding on the level of generalization vs. detail used in creating a boundary.  Generalized boundaries tend to be more fit for thematic mapping and display, where more detailed boundaries may be better for point-in-poly analysis.  Would you comment on the different criteria that is used in deciding how to generalize postal boundaries and what users may want to consider when selecting a postal boundary for different applications?  

I'm curious because I'm talking with a student that would like to conduct detailed analysis for a market study.  She has a set of GPS data collected from a mobile phone app, and she would like to ensure she is assigning the correct ZIP code to clusters of the data, so precision and accuracy are important.  However, she would then like to create images for display and I'm concerned that if she used the same, detailed boundaries for display, she is going to end up with a map that looks odd (and ugly).

------------------------------
Dan Adams
Pitney Bowes
VP, Data Product Management
White River Junction VT
------------------------------

Original Message:
Sent: 03-14-2019 12:25
From: John Tagliaferro
Subject: Post Code Boundaries vs Points


As a PM for world postcode boundaries I'm often asked do we have boundaries for all the postcodes in a given country?  Or why don't we have boundaries for 6 digit postal codes in Canada or 8 digit postal codes in Brazil?

It's important to understand a bit more about how postal code boundaries are commonly generated.  Remember first that a postal code boundary does not typically exist in reality.  Postal codes are a means of delivering mail and don't need to conform to administrative or natural boundaries.  They are created using a combination of postal delivery points and routes and building blocks of geography.  When you have enough delivery points a boundary becomes possible.

When you look at the whole of postal codes and types of delivery in a postal system however, you encounter postal delivery types that are unique businesses, post office boxes, military collection centers.   When we create postal products we commonly represent these as points, not polygons because they have specific locations.   So within our 5 digit ZIP code product in order to represent a full roster of postal codes you would be working with both points and polygons.

For postal codes like 6 digit codes in Canada (local delivery units) and ZIP +4's in the USA, these often represent street sides and multi unit dwellings. Centroids are then generated for a representative location vs attempting to render a polygon for such a localized area.

Finally, our products often include centroids pre-calculated for boundaries allowing you to have a full roster of postal codes as points.   Polygons however are typically reserved for postal codes where they can derived with the right criteria being met.

------------------------------
John Tagliaferro
Product Manager - LI Data Product Team
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

Hi Dan,

Thanks for the question.   The 5 Digit ZIP code boundaries come in both generalized and ungeneralized files.   I've taken some screenshots to show the differences.   From a cartographic display standpoint the generalization doesn't cause the boundary to take on poor display qualities until you zoom in very close.   There will be a difference however when you are dropping points along a street, especially curved streets.  That can cause a point to fall on either side of a polygon depending on which flavor you choose. The fact is that there isn't a huge difference between gen and ungen ZIP codes because while there are a number of nodes that are dropped the algorithms that are used do very well to maintain shape/integrity of the polygons.  Choosing generalized boundaries is primarily for those users who want to work with ZIP codes in a web application that needs high performance and quick redraw times.  In general you should go with the ungeneralized - they are the most detailed and still display well.

Here are few screenshots looking at ZIP codes in New Hampshire, I've added ZIP+4 centroids in a couple as well.





------------------------------
John Tagliaferro
Product Manager LI Data
PITNEY BOWES SOFTWARE, INC
White River Junction, VT

------------------------------

Original Message:
Sent: 03-19-2019 13:01
From: Dan Adams
Subject: Post Code Boundaries vs Points

John, thanks for the information on how PB represents postal boundaries and postal points.  I know one of the challenges in creating boundaries is deciding on the level of generalization vs. detail used in creating a boundary.  Generalized boundaries tend to be more fit for thematic mapping and display, where more detailed boundaries may be better for point-in-poly analysis.  Would you comment on the different criteria that is used in deciding how to generalize postal boundaries and what users may want to consider when selecting a postal boundary for different applications?

I'm curious because I'm talking with a student that would like to conduct detailed analysis for a market study.  She has a set of GPS data collected from a mobile phone app, and she would like to ensure she is assigning the correct ZIP code to clusters of the data, so precision and accuracy are important.  However, she would then like to create images for display and I'm concerned that if she used the same, detailed boundaries for display, she is going to end up with a map that looks odd (and ugly).

------------------------------
Dan Adams
Pitney Bowes
VP, Data Product Management
White River Junction VT

Original Message:
Sent: 03-14-2019 12:25
From: John Tagliaferro
Subject: Post Code Boundaries vs Points


As a PM for world postcode boundaries I'm often asked do we have boundaries for all the postcodes in a given country?  Or why don't we have boundaries for 6 digit postal codes in Canada or 8 digit postal codes in Brazil?

It's important to understand a bit more about how postal code boundaries are commonly generated.  Remember first that a postal code boundary does not typically exist in reality.  Postal codes are a means of delivering mail and don't need to conform to administrative or natural boundaries.  They are created using a combination of postal delivery points and routes and building blocks of geography.  When you have enough delivery points a boundary becomes possible.

When you look at the whole of postal codes and types of delivery in a postal system however, you encounter postal delivery types that are unique businesses, post office boxes, military collection centers.   When we create postal products we commonly represent these as points, not polygons because they have specific locations.   So within our 5 digit ZIP code product in order to represent a full roster of postal codes you would be working with both points and polygons.

For postal codes like 6 digit codes in Canada (local delivery units) and ZIP +4's in the USA, these often represent street sides and multi unit dwellings. Centroids are then generated for a representative location vs attempting to render a polygon for such a localized area.

Finally, our products often include centroids pre-calculated for boundaries allowing you to have a full roster of postal codes as points.   Polygons however are typically reserved for postal codes where they can derived with the right criteria being met.

------------------------------
John Tagliaferro
Product Manager - LI Data Product Team
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

Hi John,

With all this discussion about digit zipcodes, I thought I'd share details on the British Postcode system as we do tend to get a lot of questions around it with customers outside the UK sometimes struggling to understand the Postcode Sector, Postcode District and Postcode Area layers in our UK postcode boundary product PostMap UK.

The United Kingdom Postcode system was designed independently of any administrative boundaries so it will not align to them cleanly and may cross administrative boundaries in some cases.  The Postcode Sectors, Districts and Areas naturally align within their own geography.

The UK Postcode system starts with the Full Unit Postcode, e.g. "AB12 3DE", the postcode unit represents on average 15 households and was originally modeled on "a postman's walk".  There are over 1.6 million postcodes in the UK. It is challenging to capture the polygons at this level as with so few households within each unit postcode they can be found representing individual floors of high rise buildings and some business locations may have multiple postcodes where they are large users.  Ordnance Survey, the UK mapping agency captures this data in their Codepoint with Polygons product. 

The next level of postcodes up from Unit Level (e.g. "AB12 3DE") is the Postcode sector - in this example. "AB12 3" this represents around 2500 households and is commonly used to segment demographic information.

The next level above postcode sector is Postcode District, so taking our example of a unit postcode "AB12 3DE" , the postcode district that this unit postcode falls within is "AB12"  this represents around 8,500 households.

The final level of UK postcode boundaries is Postcode Area.  Taking our example of a unit postcode "AB12 3DE", the postcode area that this unit postcode falls within is "AB"  and this will contain approximately 190,000 households.

The above image shows how the postcode sectors, districts and areas nest within each other.

------------------------------
Daniel Edwards
Product Manager, Location Intelligence Data
Pitney Bowes Software Ltd
HENLEY-ON-THAMES
------------------------------

Original Message:
Sent: 03-21-2019 16:33
From: John Tagliaferro
Subject: Post Code Boundaries vs Points

Hi Dan,

Thanks for the question.   The 5 Digit ZIP code boundaries come in both generalized and ungeneralized files.   I've taken some screenshots to show the differences.   From a cartographic display standpoint the generalization doesn't cause the boundary to take on poor display qualities until you zoom in very close.   There will be a difference however when you are dropping points along a street, especially curved streets.  That can cause a point to fall on either side of a polygon depending on which flavor you choose. The fact is that there isn't a huge difference between gen and ungen ZIP codes because while there are a number of nodes that are dropped the algorithms that are used do very well to maintain shape/integrity of the polygons.  Choosing generalized boundaries is primarily for those users who want to work with ZIP codes in a web application that needs high performance and quick redraw times.  In general you should go with the ungeneralized - they are the most detailed and still display well.

Here are few screenshots looking at ZIP codes in New Hampshire, I've added ZIP+4 centroids in a couple as well.





------------------------------
John Tagliaferro
Product Manager LI Data
PITNEY BOWES SOFTWARE, INC
White River Junction, VT


Original Message:
Sent: 03-19-2019 13:01
From: Dan Adams
Subject: Post Code Boundaries vs Points

John, thanks for the information on how PB represents postal boundaries and postal points.  I know one of the challenges in creating boundaries is deciding on the level of generalization vs. detail used in creating a boundary.  Generalized boundaries tend to be more fit for thematic mapping and display, where more detailed boundaries may be better for point-in-poly analysis.  Would you comment on the different criteria that is used in deciding how to generalize postal boundaries and what users may want to consider when selecting a postal boundary for different applications?

I'm curious because I'm talking with a student that would like to conduct detailed analysis for a market study.  She has a set of GPS data collected from a mobile phone app, and she would like to ensure she is assigning the correct ZIP code to clusters of the data, so precision and accuracy are important.  However, she would then like to create images for display and I'm concerned that if she used the same, detailed boundaries for display, she is going to end up with a map that looks odd (and ugly).

------------------------------
Dan Adams
Pitney Bowes
VP, Data Product Management
White River Junction VT

Original Message:
Sent: 03-14-2019 12:25
From: John Tagliaferro
Subject: Post Code Boundaries vs Points


As a PM for world postcode boundaries I'm often asked do we have boundaries for all the postcodes in a given country?  Or why don't we have boundaries for 6 digit postal codes in Canada or 8 digit postal codes in Brazil?

It's important to understand a bit more about how postal code boundaries are commonly generated.  Remember first that a postal code boundary does not typically exist in reality.  Postal codes are a means of delivering mail and don't need to conform to administrative or natural boundaries.  They are created using a combination of postal delivery points and routes and building blocks of geography.  When you have enough delivery points a boundary becomes possible.

When you look at the whole of postal codes and types of delivery in a postal system however, you encounter postal delivery types that are unique businesses, post office boxes, military collection centers.   When we create postal products we commonly represent these as points, not polygons because they have specific locations.   So within our 5 digit ZIP code product in order to represent a full roster of postal codes you would be working with both points and polygons.

For postal codes like 6 digit codes in Canada (local delivery units) and ZIP +4's in the USA, these often represent street sides and multi unit dwellings. Centroids are then generated for a representative location vs attempting to render a polygon for such a localized area.

Finally, our products often include centroids pre-calculated for boundaries allowing you to have a full roster of postal codes as points.   Polygons however are typically reserved for postal codes where they can derived with the right criteria being met.

------------------------------
John Tagliaferro
Product Manager - LI Data Product Team
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

Hi Dan

Great explanation.

Do you know how many new postcodes are created annually? I guess that some get retired, split or even re-used?

------------------------------
Andy Peloe
Senior Product Manager - World Demographics
Pitney Bowes
------------------------------

Original Message:
Sent: 03-25-2019 19:03
From: Daniel Edwards
Subject: Post Code Boundaries vs Points

Hi John,

With all this discussion about digit zipcodes, I thought I'd share details on the British Postcode system as we do tend to get a lot of questions around it with customers outside the UK sometimes struggling to understand the Postcode Sector, Postcode District and Postcode Area layers in our UK postcode boundary product PostMap UK.

The United Kingdom Postcode system was designed independently of any administrative boundaries so it will not align to them cleanly and may cross administrative boundaries in some cases.  The Postcode Sectors, Districts and Areas naturally align within their own geography.

The UK Postcode system starts with the Full Unit Postcode, e.g. "AB12 3DE", the postcode unit represents on average 15 households and was originally modeled on "a postman's walk".  There are over 1.6 million postcodes in the UK. It is challenging to capture the polygons at this level as with so few households within each unit postcode they can be found representing individual floors of high rise buildings and some business locations may have multiple postcodes where they are large users.  Ordnance Survey, the UK mapping agency captures this data in their Codepoint with Polygons product.

The next level of postcodes up from Unit Level (e.g. "AB12 3DE") is the Postcode sector - in this example. "AB12 3" this represents around 2500 households and is commonly used to segment demographic information.

The next level above postcode sector is Postcode District, so taking our example of a unit postcode "AB12 3DE" , the postcode district that this unit postcode falls within is "AB12"  this represents around 8,500 households.

The final level of UK postcode boundaries is Postcode Area.  Taking our example of a unit postcode "AB12 3DE", the postcode area that this unit postcode falls within is "AB"  and this will contain approximately 190,000 households.

The above image shows how the postcode sectors, districts and areas nest within each other.

------------------------------
Daniel Edwards
Product Manager, Location Intelligence Data
Pitney Bowes Software Ltd
HENLEY-ON-THAMES

Original Message:
Sent: 03-21-2019 16:33
From: John Tagliaferro
Subject: Post Code Boundaries vs Points

Hi Dan,

Thanks for the question.   The 5 Digit ZIP code boundaries come in both generalized and ungeneralized files.   I've taken some screenshots to show the differences.   From a cartographic display standpoint the generalization doesn't cause the boundary to take on poor display qualities until you zoom in very close.   There will be a difference however when you are dropping points along a street, especially curved streets.  That can cause a point to fall on either side of a polygon depending on which flavor you choose. The fact is that there isn't a huge difference between gen and ungen ZIP codes because while there are a number of nodes that are dropped the algorithms that are used do very well to maintain shape/integrity of the polygons.  Choosing generalized boundaries is primarily for those users who want to work with ZIP codes in a web application that needs high performance and quick redraw times.  In general you should go with the ungeneralized - they are the most detailed and still display well.

Here are few screenshots looking at ZIP codes in New Hampshire, I've added ZIP+4 centroids in a couple as well.





------------------------------
John Tagliaferro
Product Manager LI Data
PITNEY BOWES SOFTWARE, INC
White River Junction, VT


Original Message:
Sent: 03-19-2019 13:01
From: Dan Adams
Subject: Post Code Boundaries vs Points

John, thanks for the information on how PB represents postal boundaries and postal points.  I know one of the challenges in creating boundaries is deciding on the level of generalization vs. detail used in creating a boundary.  Generalized boundaries tend to be more fit for thematic mapping and display, where more detailed boundaries may be better for point-in-poly analysis.  Would you comment on the different criteria that is used in deciding how to generalize postal boundaries and what users may want to consider when selecting a postal boundary for different applications?

I'm curious because I'm talking with a student that would like to conduct detailed analysis for a market study.  She has a set of GPS data collected from a mobile phone app, and she would like to ensure she is assigning the correct ZIP code to clusters of the data, so precision and accuracy are important.  However, she would then like to create images for display and I'm concerned that if she used the same, detailed boundaries for display, she is going to end up with a map that looks odd (and ugly).

------------------------------
Dan Adams
Pitney Bowes
VP, Data Product Management
White River Junction VT

Original Message:
Sent: 03-14-2019 12:25
From: John Tagliaferro
Subject: Post Code Boundaries vs Points


As a PM for world postcode boundaries I'm often asked do we have boundaries for all the postcodes in a given country?  Or why don't we have boundaries for 6 digit postal codes in Canada or 8 digit postal codes in Brazil?

It's important to understand a bit more about how postal code boundaries are commonly generated.  Remember first that a postal code boundary does not typically exist in reality.  Postal codes are a means of delivering mail and don't need to conform to administrative or natural boundaries.  They are created using a combination of postal delivery points and routes and building blocks of geography.  When you have enough delivery points a boundary becomes possible.

When you look at the whole of postal codes and types of delivery in a postal system however, you encounter postal delivery types that are unique businesses, post office boxes, military collection centers.   When we create postal products we commonly represent these as points, not polygons because they have specific locations.   So within our 5 digit ZIP code product in order to represent a full roster of postal codes you would be working with both points and polygons.

For postal codes like 6 digit codes in Canada (local delivery units) and ZIP +4's in the USA, these often represent street sides and multi unit dwellings. Centroids are then generated for a representative location vs attempting to render a polygon for such a localized area.

Finally, our products often include centroids pre-calculated for boundaries allowing you to have a full roster of postal codes as points.   Polygons however are typically reserved for postal codes where they can derived with the right criteria being met.

------------------------------
John Tagliaferro
Product Manager - LI Data Product Team
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

Hi Andy,

Changes to the number of Postcode Areas and Districts are rare.  We get more changes in the Postcode Sectors.  For example in Postmap release 67 (which is our current version as I write this) we saw 11 new Postcode Sectors added such as CB3 1 in Cambridge.  We also saw 2 sectors removed and several boundary realignments.

In terms of Unit Postcodes, it naturally fluctuates, looking at our Postpoint Professional product (Postcode points) which is released every 6 months.  In the 6 months to January 2019 we saw an overall increase in the number of postcodes of 1342 up to 1706504 for Great Britain but delving deeper into the data we actually see that there were 4807 new unit postcodes and 3465 deleted postcodes.  For Northern Ireland we see 48144 postcodes which is an overall increase of 31 but again that actually represents 80 new and 49 deleted postcodes.

So a relatively quiet period of postcode turnover.  Compare this to the first half of 2018 and we see over 4000 postcodes added to Great Britain of which there was over 9500 new postcodes added and over 5000 removed.


------------------------------
Daniel Edwards
Product Manager, Location Intelligence Data
Pitney Bowes Software Ltd
HENLEY-ON-THAMES
------------------------------

Original Message:
Sent: 03-29-2019 09:55
From: Andrew Peloe
Subject: Post Code Boundaries vs Points

Hi Dan

Great explanation.

Do you know how many new postcodes are created annually? I guess that some get retired, split or even re-used?

------------------------------
Andy Peloe
Senior Product Manager - World Demographics
Pitney Bowes

Original Message:
Sent: 03-25-2019 19:03
From: Daniel Edwards
Subject: Post Code Boundaries vs Points

Hi John,

With all this discussion about digit zipcodes, I thought I'd share details on the British Postcode system as we do tend to get a lot of questions around it with customers outside the UK sometimes struggling to understand the Postcode Sector, Postcode District and Postcode Area layers in our UK postcode boundary product PostMap UK.

The United Kingdom Postcode system was designed independently of any administrative boundaries so it will not align to them cleanly and may cross administrative boundaries in some cases.  The Postcode Sectors, Districts and Areas naturally align within their own geography.

The UK Postcode system starts with the Full Unit Postcode, e.g. "AB12 3DE", the postcode unit represents on average 15 households and was originally modeled on "a postman's walk".  There are over 1.6 million postcodes in the UK. It is challenging to capture the polygons at this level as with so few households within each unit postcode they can be found representing individual floors of high rise buildings and some business locations may have multiple postcodes where they are large users.  Ordnance Survey, the UK mapping agency captures this data in their Codepoint with Polygons product.

The next level of postcodes up from Unit Level (e.g. "AB12 3DE") is the Postcode sector - in this example. "AB12 3" this represents around 2500 households and is commonly used to segment demographic information.

The next level above postcode sector is Postcode District, so taking our example of a unit postcode "AB12 3DE" , the postcode district that this unit postcode falls within is "AB12"  this represents around 8,500 households.

The final level of UK postcode boundaries is Postcode Area.  Taking our example of a unit postcode "AB12 3DE", the postcode area that this unit postcode falls within is "AB"  and this will contain approximately 190,000 households.

The above image shows how the postcode sectors, districts and areas nest within each other.

------------------------------
Daniel Edwards
Product Manager, Location Intelligence Data
Pitney Bowes Software Ltd
HENLEY-ON-THAMES

Original Message:
Sent: 03-21-2019 16:33
From: John Tagliaferro
Subject: Post Code Boundaries vs Points

Hi Dan,

Thanks for the question.   The 5 Digit ZIP code boundaries come in both generalized and ungeneralized files.   I've taken some screenshots to show the differences.   From a cartographic display standpoint the generalization doesn't cause the boundary to take on poor display qualities until you zoom in very close.   There will be a difference however when you are dropping points along a street, especially curved streets.  That can cause a point to fall on either side of a polygon depending on which flavor you choose. The fact is that there isn't a huge difference between gen and ungen ZIP codes because while there are a number of nodes that are dropped the algorithms that are used do very well to maintain shape/integrity of the polygons.  Choosing generalized boundaries is primarily for those users who want to work with ZIP codes in a web application that needs high performance and quick redraw times.  In general you should go with the ungeneralized - they are the most detailed and still display well.

Here are few screenshots looking at ZIP codes in New Hampshire, I've added ZIP+4 centroids in a couple as well.





------------------------------
John Tagliaferro
Product Manager LI Data
PITNEY BOWES SOFTWARE, INC
White River Junction, VT


Original Message:
Sent: 03-19-2019 13:01
From: Dan Adams
Subject: Post Code Boundaries vs Points

John, thanks for the information on how PB represents postal boundaries and postal points.  I know one of the challenges in creating boundaries is deciding on the level of generalization vs. detail used in creating a boundary.  Generalized boundaries tend to be more fit for thematic mapping and display, where more detailed boundaries may be better for point-in-poly analysis.  Would you comment on the different criteria that is used in deciding how to generalize postal boundaries and what users may want to consider when selecting a postal boundary for different applications?

I'm curious because I'm talking with a student that would like to conduct detailed analysis for a market study.  She has a set of GPS data collected from a mobile phone app, and she would like to ensure she is assigning the correct ZIP code to clusters of the data, so precision and accuracy are important.  However, she would then like to create images for display and I'm concerned that if she used the same, detailed boundaries for display, she is going to end up with a map that looks odd (and ugly).

------------------------------
Dan Adams
Pitney Bowes
VP, Data Product Management
White River Junction VT

Original Message:
Sent: 03-14-2019 12:25
From: John Tagliaferro
Subject: Post Code Boundaries vs Points


As a PM for world postcode boundaries I'm often asked do we have boundaries for all the postcodes in a given country?  Or why don't we have boundaries for 6 digit postal codes in Canada or 8 digit postal codes in Brazil?

It's important to understand a bit more about how postal code boundaries are commonly generated.  Remember first that a postal code boundary does not typically exist in reality.  Postal codes are a means of delivering mail and don't need to conform to administrative or natural boundaries.  They are created using a combination of postal delivery points and routes and building blocks of geography.  When you have enough delivery points a boundary becomes possible.

When you look at the whole of postal codes and types of delivery in a postal system however, you encounter postal delivery types that are unique businesses, post office boxes, military collection centers.   When we create postal products we commonly represent these as points, not polygons because they have specific locations.   So within our 5 digit ZIP code product in order to represent a full roster of postal codes you would be working with both points and polygons.

For postal codes like 6 digit codes in Canada (local delivery units) and ZIP +4's in the USA, these often represent street sides and multi unit dwellings. Centroids are then generated for a representative location vs attempting to render a polygon for such a localized area.

Finally, our products often include centroids pre-calculated for boundaries allowing you to have a full roster of postal codes as points.   Polygons however are typically reserved for postal codes where they can derived with the right criteria being met.

------------------------------
John Tagliaferro
Product Manager - LI Data Product Team
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

Hello Dan,
   Are the Postcode changes typically spread across the country or are there parts of the United Kingdom that are really seeing a lot of population growth thus most of the changes are located there?  I was curious about this in regard to automated quality checks.  If historically the change has been evenly spread around or localized around a few urban areas then we could construct a quality check to look for abnormal amounts of change.




------------------------------
Jeffrey Howe
Quality Assurance Lead
Pitney Bowes
White River Junction VT
------------------------------

Original Message:
Sent: 04-05-2019 12:45
From: Daniel Edwards
Subject: Post Code Boundaries vs Points

Hi Andy,

Changes to the number of Postcode Areas and Districts are rare.  We get more changes in the Postcode Sectors.  For example in Postmap release 67 (which is our current version as I write this) we saw 11 new Postcode Sectors added such as CB3 1 in Cambridge.  We also saw 2 sectors removed and several boundary realignments.

In terms of Unit Postcodes, it naturally fluctuates, looking at our Postpoint Professional product (Postcode points) which is released every 6 months.  In the 6 months to January 2019 we saw an overall increase in the number of postcodes of 1342 up to 1706504 for Great Britain but delving deeper into the data we actually see that there were 4807 new unit postcodes and 3465 deleted postcodes.  For Northern Ireland we see 48144 postcodes which is an overall increase of 31 but again that actually represents 80 new and 49 deleted postcodes.

So a relatively quiet period of postcode turnover.  Compare this to the first half of 2018 and we see over 4000 postcodes added to Great Britain of which there was over 9500 new postcodes added and over 5000 removed.


------------------------------
Daniel Edwards
Product Manager, Location Intelligence Data
Pitney Bowes Software Ltd
HENLEY-ON-THAMES

Original Message:
Sent: 03-29-2019 09:55
From: Andrew Peloe
Subject: Post Code Boundaries vs Points

Hi Dan

Great explanation.

Do you know how many new postcodes are created annually? I guess that some get retired, split or even re-used?

------------------------------
Andy Peloe
Senior Product Manager - World Demographics
Pitney Bowes

Original Message:
Sent: 03-25-2019 19:03
From: Daniel Edwards
Subject: Post Code Boundaries vs Points

Hi John,

With all this discussion about digit zipcodes, I thought I'd share details on the British Postcode system as we do tend to get a lot of questions around it with customers outside the UK sometimes struggling to understand the Postcode Sector, Postcode District and Postcode Area layers in our UK postcode boundary product PostMap UK.

The United Kingdom Postcode system was designed independently of any administrative boundaries so it will not align to them cleanly and may cross administrative boundaries in some cases.  The Postcode Sectors, Districts and Areas naturally align within their own geography.

The UK Postcode system starts with the Full Unit Postcode, e.g. "AB12 3DE", the postcode unit represents on average 15 households and was originally modeled on "a postman's walk".  There are over 1.6 million postcodes in the UK. It is challenging to capture the polygons at this level as with so few households within each unit postcode they can be found representing individual floors of high rise buildings and some business locations may have multiple postcodes where they are large users.  Ordnance Survey, the UK mapping agency captures this data in their Codepoint with Polygons product.

The next level of postcodes up from Unit Level (e.g. "AB12 3DE") is the Postcode sector - in this example. "AB12 3" this represents around 2500 households and is commonly used to segment demographic information.

The next level above postcode sector is Postcode District, so taking our example of a unit postcode "AB12 3DE" , the postcode district that this unit postcode falls within is "AB12"  this represents around 8,500 households.

The final level of UK postcode boundaries is Postcode Area.  Taking our example of a unit postcode "AB12 3DE", the postcode area that this unit postcode falls within is "AB"  and this will contain approximately 190,000 households.

The above image shows how the postcode sectors, districts and areas nest within each other.

------------------------------
Daniel Edwards
Product Manager, Location Intelligence Data
Pitney Bowes Software Ltd
HENLEY-ON-THAMES

Original Message:
Sent: 03-21-2019 16:33
From: John Tagliaferro
Subject: Post Code Boundaries vs Points

Hi Dan,

Thanks for the question.   The 5 Digit ZIP code boundaries come in both generalized and ungeneralized files.   I've taken some screenshots to show the differences.   From a cartographic display standpoint the generalization doesn't cause the boundary to take on poor display qualities until you zoom in very close.   There will be a difference however when you are dropping points along a street, especially curved streets.  That can cause a point to fall on either side of a polygon depending on which flavor you choose. The fact is that there isn't a huge difference between gen and ungen ZIP codes because while there are a number of nodes that are dropped the algorithms that are used do very well to maintain shape/integrity of the polygons.  Choosing generalized boundaries is primarily for those users who want to work with ZIP codes in a web application that needs high performance and quick redraw times.  In general you should go with the ungeneralized - they are the most detailed and still display well.

Here are few screenshots looking at ZIP codes in New Hampshire, I've added ZIP+4 centroids in a couple as well.





------------------------------
John Tagliaferro
Product Manager LI Data
PITNEY BOWES SOFTWARE, INC
White River Junction, VT


Original Message:
Sent: 03-19-2019 13:01
From: Dan Adams
Subject: Post Code Boundaries vs Points

John, thanks for the information on how PB represents postal boundaries and postal points.  I know one of the challenges in creating boundaries is deciding on the level of generalization vs. detail used in creating a boundary.  Generalized boundaries tend to be more fit for thematic mapping and display, where more detailed boundaries may be better for point-in-poly analysis.  Would you comment on the different criteria that is used in deciding how to generalize postal boundaries and what users may want to consider when selecting a postal boundary for different applications?

I'm curious because I'm talking with a student that would like to conduct detailed analysis for a market study.  She has a set of GPS data collected from a mobile phone app, and she would like to ensure she is assigning the correct ZIP code to clusters of the data, so precision and accuracy are important.  However, she would then like to create images for display and I'm concerned that if she used the same, detailed boundaries for display, she is going to end up with a map that looks odd (and ugly).

------------------------------
Dan Adams
Pitney Bowes
VP, Data Product Management
White River Junction VT

Original Message:
Sent: 03-14-2019 12:25
From: John Tagliaferro
Subject: Post Code Boundaries vs Points


As a PM for world postcode boundaries I'm often asked do we have boundaries for all the postcodes in a given country?  Or why don't we have boundaries for 6 digit postal codes in Canada or 8 digit postal codes in Brazil?

It's important to understand a bit more about how postal code boundaries are commonly generated.  Remember first that a postal code boundary does not typically exist in reality.  Postal codes are a means of delivering mail and don't need to conform to administrative or natural boundaries.  They are created using a combination of postal delivery points and routes and building blocks of geography.  When you have enough delivery points a boundary becomes possible.

When you look at the whole of postal codes and types of delivery in a postal system however, you encounter postal delivery types that are unique businesses, post office boxes, military collection centers.   When we create postal products we commonly represent these as points, not polygons because they have specific locations.   So within our 5 digit ZIP code product in order to represent a full roster of postal codes you would be working with both points and polygons.

For postal codes like 6 digit codes in Canada (local delivery units) and ZIP +4's in the USA, these often represent street sides and multi unit dwellings. Centroids are then generated for a representative location vs attempting to render a polygon for such a localized area.

Finally, our products often include centroids pre-calculated for boundaries allowing you to have a full roster of postal codes as points.   Polygons however are typically reserved for postal codes where they can derived with the right criteria being met.

------------------------------
John Tagliaferro
Product Manager - LI Data Product Team
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

Hi Jeff,

Postcode additions and deletions tend to occur most where buildings are being built, redeveloped and demolished. These tend to be most common in built up, populated areas. There will also be some turn over of businesses who get a sufficient volume of mail to warrant a postcode being given for their usage.

I was looking at how best to visualize this and came up with the following two animations below. They show the accumulated postcode adds and deletes over the space of a year. The concentrations mirror pretty much the major population centers of the UK


------------------------------
James Morgan
UK Data Development Manager
Pitney Bowes
Henley-on-Thames

------------------------------

Original Message:
Sent: 04-08-2019 10:20
From: Jeffrey Howe
Subject: Post Code Boundaries vs Points

Hello Dan,
   Are the Postcode changes typically spread across the country or are there parts of the United Kingdom that are really seeing a lot of population growth thus most of the changes are located there?  I was curious about this in regard to automated quality checks.  If historically the change has been evenly spread around or localized around a few urban areas then we could construct a quality check to look for abnormal amounts of change.




------------------------------
Jeffrey Howe
Quality Assurance Lead
Pitney Bowes
White River Junction VT

Original Message:
Sent: 04-05-2019 12:45
From: Daniel Edwards
Subject: Post Code Boundaries vs Points

Hi Andy,

Changes to the number of Postcode Areas and Districts are rare.  We get more changes in the Postcode Sectors.  For example in Postmap release 67 (which is our current version as I write this) we saw 11 new Postcode Sectors added such as CB3 1 in Cambridge.  We also saw 2 sectors removed and several boundary realignments.

In terms of Unit Postcodes, it naturally fluctuates, looking at our Postpoint Professional product (Postcode points) which is released every 6 months.  In the 6 months to January 2019 we saw an overall increase in the number of postcodes of 1342 up to 1706504 for Great Britain but delving deeper into the data we actually see that there were 4807 new unit postcodes and 3465 deleted postcodes.  For Northern Ireland we see 48144 postcodes which is an overall increase of 31 but again that actually represents 80 new and 49 deleted postcodes.

So a relatively quiet period of postcode turnover.  Compare this to the first half of 2018 and we see over 4000 postcodes added to Great Britain of which there was over 9500 new postcodes added and over 5000 removed.


------------------------------
Daniel Edwards
Product Manager, Location Intelligence Data
Pitney Bowes Software Ltd
HENLEY-ON-THAMES

Original Message:
Sent: 03-29-2019 09:55
From: Andrew Peloe
Subject: Post Code Boundaries vs Points

Hi Dan

Great explanation.

Do you know how many new postcodes are created annually? I guess that some get retired, split or even re-used?

------------------------------
Andy Peloe
Senior Product Manager - World Demographics
Pitney Bowes

Original Message:
Sent: 03-25-2019 19:03
From: Daniel Edwards
Subject: Post Code Boundaries vs Points

Hi John,

With all this discussion about digit zipcodes, I thought I'd share details on the British Postcode system as we do tend to get a lot of questions around it with customers outside the UK sometimes struggling to understand the Postcode Sector, Postcode District and Postcode Area layers in our UK postcode boundary product PostMap UK.

The United Kingdom Postcode system was designed independently of any administrative boundaries so it will not align to them cleanly and may cross administrative boundaries in some cases.  The Postcode Sectors, Districts and Areas naturally align within their own geography.

The UK Postcode system starts with the Full Unit Postcode, e.g. "AB12 3DE", the postcode unit represents on average 15 households and was originally modeled on "a postman's walk".  There are over 1.6 million postcodes in the UK. It is challenging to capture the polygons at this level as with so few households within each unit postcode they can be found representing individual floors of high rise buildings and some business locations may have multiple postcodes where they are large users.  Ordnance Survey, the UK mapping agency captures this data in their Codepoint with Polygons product.

The next level of postcodes up from Unit Level (e.g. "AB12 3DE") is the Postcode sector - in this example. "AB12 3" this represents around 2500 households and is commonly used to segment demographic information.

The next level above postcode sector is Postcode District, so taking our example of a unit postcode "AB12 3DE" , the postcode district that this unit postcode falls within is "AB12"  this represents around 8,500 households.

The final level of UK postcode boundaries is Postcode Area.  Taking our example of a unit postcode "AB12 3DE", the postcode area that this unit postcode falls within is "AB"  and this will contain approximately 190,000 households.

The above image shows how the postcode sectors, districts and areas nest within each other.

------------------------------
Daniel Edwards
Product Manager, Location Intelligence Data
Pitney Bowes Software Ltd
HENLEY-ON-THAMES

Original Message:
Sent: 03-21-2019 16:33
From: John Tagliaferro
Subject: Post Code Boundaries vs Points

Hi Dan,

Thanks for the question.   The 5 Digit ZIP code boundaries come in both generalized and ungeneralized files.   I've taken some screenshots to show the differences.   From a cartographic display standpoint the generalization doesn't cause the boundary to take on poor display qualities until you zoom in very close.   There will be a difference however when you are dropping points along a street, especially curved streets.  That can cause a point to fall on either side of a polygon depending on which flavor you choose. The fact is that there isn't a huge difference between gen and ungen ZIP codes because while there are a number of nodes that are dropped the algorithms that are used do very well to maintain shape/integrity of the polygons.  Choosing generalized boundaries is primarily for those users who want to work with ZIP codes in a web application that needs high performance and quick redraw times.  In general you should go with the ungeneralized - they are the most detailed and still display well.

Here are few screenshots looking at ZIP codes in New Hampshire, I've added ZIP+4 centroids in a couple as well.





------------------------------
John Tagliaferro
Product Manager LI Data
PITNEY BOWES SOFTWARE, INC
White River Junction, VT


Original Message:
Sent: 03-19-2019 13:01
From: Dan Adams
Subject: Post Code Boundaries vs Points

John, thanks for the information on how PB represents postal boundaries and postal points.  I know one of the challenges in creating boundaries is deciding on the level of generalization vs. detail used in creating a boundary.  Generalized boundaries tend to be more fit for thematic mapping and display, where more detailed boundaries may be better for point-in-poly analysis.  Would you comment on the different criteria that is used in deciding how to generalize postal boundaries and what users may want to consider when selecting a postal boundary for different applications?

I'm curious because I'm talking with a student that would like to conduct detailed analysis for a market study.  She has a set of GPS data collected from a mobile phone app, and she would like to ensure she is assigning the correct ZIP code to clusters of the data, so precision and accuracy are important.  However, she would then like to create images for display and I'm concerned that if she used the same, detailed boundaries for display, she is going to end up with a map that looks odd (and ugly).

------------------------------
Dan Adams
Pitney Bowes
VP, Data Product Management
White River Junction VT

Original Message:
Sent: 03-14-2019 12:25
From: John Tagliaferro
Subject: Post Code Boundaries vs Points


As a PM for world postcode boundaries I'm often asked do we have boundaries for all the postcodes in a given country?  Or why don't we have boundaries for 6 digit postal codes in Canada or 8 digit postal codes in Brazil?

It's important to understand a bit more about how postal code boundaries are commonly generated.  Remember first that a postal code boundary does not typically exist in reality.  Postal codes are a means of delivering mail and don't need to conform to administrative or natural boundaries.  They are created using a combination of postal delivery points and routes and building blocks of geography.  When you have enough delivery points a boundary becomes possible.

When you look at the whole of postal codes and types of delivery in a postal system however, you encounter postal delivery types that are unique businesses, post office boxes, military collection centers.   When we create postal products we commonly represent these as points, not polygons because they have specific locations.   So within our 5 digit ZIP code product in order to represent a full roster of postal codes you would be working with both points and polygons.

For postal codes like 6 digit codes in Canada (local delivery units) and ZIP +4's in the USA, these often represent street sides and multi unit dwellings. Centroids are then generated for a representative location vs attempting to render a polygon for such a localized area.

Finally, our products often include centroids pre-calculated for boundaries allowing you to have a full roster of postal codes as points.   Polygons however are typically reserved for postal codes where they can derived with the right criteria being met.

------------------------------
John Tagliaferro
Product Manager - LI Data Product Team
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------

Thank you <g class="gr_ gr_15 gr-alert gr_gramm gr_inline_cards gr_run_anim Punctuation replaceWithoutSep" id="15" data-gr-id="15">James</g> for the information.  The maps are a great visual to understand the change happening.

------------------------------
Jeffrey Howe
Quality Assurance Lead
Pitney Bowes
White River Junction VT
------------------------------

Original Message:
Sent: 04-16-2019 04:15
From: James Morgan
Subject: Post Code Boundaries vs Points

Hi Jeff,

Postcode additions and deletions tend to occur most where buildings are being built, redeveloped and demolished. These tend to be most common in built up, populated areas. There will also be some turn over of businesses who get a sufficient volume of mail to warrant a postcode being given for their usage.

I was looking at how best to visualize this and came up with the following two animations below. They show the accumulated postcode adds and deletes over the space of a year. The concentrations mirror pretty much the major population centers of the UK

------------------------------
James Morgan
UK Data Development Manager
Pitney Bowes
Henley-on-Thames


Original Message:
Sent: 04-08-2019 10:20
From: Jeffrey Howe
Subject: Post Code Boundaries vs Points

Hello Dan,
   Are the Postcode changes typically spread across the country or are there parts of the United Kingdom that are really seeing a lot of population growth thus most of the changes are located there?  I was curious about this in regard to automated quality checks.  If historically the change has been evenly spread around or localized around a few urban areas then we could construct a quality check to look for abnormal amounts of change.




------------------------------
Jeffrey Howe
Quality Assurance Lead
Pitney Bowes
White River Junction VT

Original Message:
Sent: 04-05-2019 12:45
From: Daniel Edwards
Subject: Post Code Boundaries vs Points

Hi Andy,

Changes to the number of Postcode Areas and Districts are rare.  We get more changes in the Postcode Sectors.  For example in Postmap release 67 (which is our current version as I write this) we saw 11 new Postcode Sectors added such as CB3 1 in Cambridge.  We also saw 2 sectors removed and several boundary realignments.

In terms of Unit Postcodes, it naturally fluctuates, looking at our Postpoint Professional product (Postcode points) which is released every 6 months.  In the 6 months to January 2019 we saw an overall increase in the number of postcodes of 1342 up to 1706504 for Great Britain but delving deeper into the data we actually see that there were 4807 new unit postcodes and 3465 deleted postcodes.  For Northern Ireland we see 48144 postcodes which is an overall increase of 31 but again that actually represents 80 new and 49 deleted postcodes.

So a relatively quiet period of postcode turnover.  Compare this to the first half of 2018 and we see over 4000 postcodes added to Great Britain of which there was over 9500 new postcodes added and over 5000 removed.


------------------------------
Daniel Edwards
Product Manager, Location Intelligence Data
Pitney Bowes Software Ltd
HENLEY-ON-THAMES

Original Message:
Sent: 03-29-2019 09:55
From: Andrew Peloe
Subject: Post Code Boundaries vs Points

Hi Dan

Great explanation.

Do you know how many new postcodes are created annually? I guess that some get retired, split or even re-used?

------------------------------
Andy Peloe
Senior Product Manager - World Demographics
Pitney Bowes

Original Message:
Sent: 03-25-2019 19:03
From: Daniel Edwards
Subject: Post Code Boundaries vs Points

Hi John,

With all this discussion about digit zipcodes, I thought I'd share details on the British Postcode system as we do tend to get a lot of questions around it with customers outside the UK sometimes struggling to understand the Postcode Sector, Postcode District and Postcode Area layers in our UK postcode boundary product PostMap UK.

The United Kingdom Postcode system was designed independently of any administrative boundaries so it will not align to them cleanly and may cross administrative boundaries in some cases.  The Postcode Sectors, Districts and Areas naturally align within their own geography.

The UK Postcode system starts with the Full Unit Postcode, e.g. "AB12 3DE", the postcode unit represents on average 15 households and was originally modeled on "a postman's walk".  There are over 1.6 million postcodes in the UK. It is challenging to capture the polygons at this level as with so few households within each unit postcode they can be found representing individual floors of high rise buildings and some business locations may have multiple postcodes where they are large users.  Ordnance Survey, the UK mapping agency captures this data in their Codepoint with Polygons product.

The next level of postcodes up from Unit Level (e.g. "AB12 3DE") is the Postcode sector - in this example. "AB12 3" this represents around 2500 households and is commonly used to segment demographic information.

The next level above postcode sector is Postcode District, so taking our example of a unit postcode "AB12 3DE" , the postcode district that this unit postcode falls within is "AB12"  this represents around 8,500 households.

The final level of UK postcode boundaries is Postcode Area.  Taking our example of a unit postcode "AB12 3DE", the postcode area that this unit postcode falls within is "AB"  and this will contain approximately 190,000 households.

The above image shows how the postcode sectors, districts and areas nest within each other.

------------------------------
Daniel Edwards
Product Manager, Location Intelligence Data
Pitney Bowes Software Ltd
HENLEY-ON-THAMES

Original Message:
Sent: 03-21-2019 16:33
From: John Tagliaferro
Subject: Post Code Boundaries vs Points

Hi Dan,

Thanks for the question.   The 5 Digit ZIP code boundaries come in both generalized and ungeneralized files.   I've taken some screenshots to show the differences.   From a cartographic display standpoint the generalization doesn't cause the boundary to take on poor display qualities until you zoom in very close.   There will be a difference however when you are dropping points along a street, especially curved streets.  That can cause a point to fall on either side of a polygon depending on which flavor you choose. The fact is that there isn't a huge difference between gen and ungen ZIP codes because while there are a number of nodes that are dropped the algorithms that are used do very well to maintain shape/integrity of the polygons.  Choosing generalized boundaries is primarily for those users who want to work with ZIP codes in a web application that needs high performance and quick redraw times.  In general you should go with the ungeneralized - they are the most detailed and still display well.

Here are few screenshots looking at ZIP codes in New Hampshire, I've added ZIP+4 centroids in a couple as well.





------------------------------
John Tagliaferro
Product Manager LI Data
PITNEY BOWES SOFTWARE, INC
White River Junction, VT


Original Message:
Sent: 03-19-2019 13:01
From: Dan Adams
Subject: Post Code Boundaries vs Points

John, thanks for the information on how PB represents postal boundaries and postal points.  I know one of the challenges in creating boundaries is deciding on the level of generalization vs. detail used in creating a boundary.  Generalized boundaries tend to be more fit for thematic mapping and display, where more detailed boundaries may be better for point-in-poly analysis.  Would you comment on the different criteria that is used in deciding how to generalize postal boundaries and what users may want to consider when selecting a postal boundary for different applications?

I'm curious because I'm talking with a student that would like to conduct detailed analysis for a market study.  She has a set of GPS data collected from a mobile phone app, and she would like to ensure she is assigning the correct ZIP code to clusters of the data, so precision and accuracy are important.  However, she would then like to create images for display and I'm concerned that if she used the same, detailed boundaries for display, she is going to end up with a map that looks odd (and ugly).

------------------------------
Dan Adams
Pitney Bowes
VP, Data Product Management
White River Junction VT

Original Message:
Sent: 03-14-2019 12:25
From: John Tagliaferro
Subject: Post Code Boundaries vs Points


As a PM for world postcode boundaries I'm often asked do we have boundaries for all the postcodes in a given country?  Or why don't we have boundaries for 6 digit postal codes in Canada or 8 digit postal codes in Brazil?

It's important to understand a bit more about how postal code boundaries are commonly generated.  Remember first that a postal code boundary does not typically exist in reality.  Postal codes are a means of delivering mail and don't need to conform to administrative or natural boundaries.  They are created using a combination of postal delivery points and routes and building blocks of geography.  When you have enough delivery points a boundary becomes possible.

When you look at the whole of postal codes and types of delivery in a postal system however, you encounter postal delivery types that are unique businesses, post office boxes, military collection centers.   When we create postal products we commonly represent these as points, not polygons because they have specific locations.   So within our 5 digit ZIP code product in order to represent a full roster of postal codes you would be working with both points and polygons.

For postal codes like 6 digit codes in Canada (local delivery units) and ZIP +4's in the USA, these often represent street sides and multi unit dwellings. Centroids are then generated for a representative location vs attempting to render a polygon for such a localized area.

Finally, our products often include centroids pre-calculated for boundaries allowing you to have a full roster of postal codes as points.   Polygons however are typically reserved for postal codes where they can derived with the right criteria being met.

------------------------------
John Tagliaferro
Product Manager - LI Data Product Team
PITNEY BOWES SOFTWARE, INC
White River Junction, VT
------------------------------