Data measures and data sources
Last updated
Was this helpful?
Last updated
Was this helpful?
Was this helpful?
Supplier data is supplied to PointX in a wide variety of formats. Some of the data is supplied without grid references and specific attribution, therefore efforts are made to supply this data. The success of this identification is measured by the accuracy flags such as positional accuracy and match indicators.
Ordnance Survey data and specialised geocoding software is used to achieve positional accuracy and to provide a confidence level indicator for each record. Data that cannot be given coordinates due to insufficient location or address detail are not supplied in the Product. It is very important that customers take the positional accuracy value of the feature into account when using it in applications.
The positional accuracy falls into one of the following categories:
Positioned to the address or location, usually within the building footprint.
Positioned to an adjacent address or location for non-addressable features.
Positioned to the road within the address or location.
Positioned within the geographical locality.
Typically, most records are positioned on or adjacent to the address or location of the feature. A very small proportion of records are positioned to the road within the address or location, or within the geographic locality of the feature. A positional accuracy code of 3 or 4 is used only for ATMs (cash machines), Wi-Fi hotspots and public telephones, where the data is sourced from a single authoritative and non-duplicated source. Continuous efforts are made to improve the accuracy of the geocoding, as shown in the table below showing the range of positional accuracy codes.
1
71.75
79.87
86.22
2
27.21
19.28
13.66
3
0.84
0.70
0.07
4
0.20
0.15
0.05
Total
100.00
100.00
100.00
If the positional accuracy is stated as ‘1’, the coordinates should fall within the footprint of the real- world feature in question, typically a building or structure.
Where the positional accuracy is stated as ‘2’, the coordinates will either be placed centrally in the text that relates to the feature – and may be within the extent of the feature – or will be positioned close to the true location of a part of the feature. In practice, it will probably be within 10 metres of either the feature or an edge of the feature’s geographic extent, although this cannot be guaranteed.
A positional accuracy of ‘3’ indicates that the coordinates are placed centrally on the correct road. As road lengths vary and the central position could be very near or a long way from the true location on the road, it is difficult to say exactly how far these instances might be shown from their true location. In the majority, it is to be expected that the assigned coordinate could be up to a kilometre away, with a small number being even farther.
A positional accuracy of ‘4’ means that the location assigned is in the correct geographic locality, such as the right village or industrial estate, and that it has not been possible to locate to a specific relevant road. Depending on how big the geographic locality in question is, the record’s true location could in reality be up to a few kilometres from the reported coordinates.
Each Points of Interest feature is matched to the closest road by performing an attribution and spatial match between the feature and OS MasterMap Highways Network. This results in each feature being given a Unique Street Reference Number (USRN). The match accuracy between the Points of Interest feature and the OS MasterMap Highways Network is determined from the USRN Match Indicator attribute. It is again very important that customers take this match indicator value of the feature into account when using it in applications.
If the match indicator is stated as ‘1’ the Points of Interest feature has been matched against the OS MasterMap Highways Network using the Unique Property Reference Number (UPRN) from AddressBase and the USRN.
Where the match indicator is stated as ‘2’, the feature has been matched spatially to the road which appears in either the address detail or street name attributes. For more information on these individual attributes please see the Technical Specification.
A match indicator of ‘3’ indicates that the USRN has been spatially matched to the nearest road in the OS MasterMap Highways Network.
Please be aware that a match indicator of 3 may not necessarily be the access road to the feature.
The data contains both addressable and non-addressable records. Structured address information is only provided for those records that are addressable and can be successfully matched against AddressBase Plus.
AddressBase Plus contains current properties using addresses sourced from Local Authorities, Ordnance Survey and Royal Mail for England, Wales and Scotland. The Unique Property Reference Number (UPRN) is the unique identifier for every addressable location in Great Britain.
Records that have been successfully matched will have a UPRN. Attempts have been made to provide address information where possible when not matched against AddressBase Plus.
Records have been attributed with both a TOID and TOID version that have been directly derived from area features within OS MasterMap Topography Layer based on the grid reference of the feature. There will be a number of features where a TOID value will not be provided because the feature does not sit within the correct OS MasterMap Topography Layer polygon, and these will have a TOID value of Not Assigned and a TOID version of ‘0’.
There are two main sources of data that provide approximately 80% of the Points of Interest features. The authoritative source for a type of feature is used to build the dataset where possible. Approximately
140 suppliers are used to create the dataset. The Points of Interest Provenance page lists all the suppliers and indicates their level of contribution to the dataset.
PointX relies upon approximately 140 data suppliers to provide the initial classification of each of the records. This is then used to allocate the appropriate class to the feature which can lead to examples where different classes are applied to the same real-world feature, dependent upon the base classification used by the different data suppliers. Part of the production process for the current product involves the removal of multiple instances of what is believed to be the same feature, it is possible that individual instances of what are, in fact, the same type of feature can be reported in different classes.