Process Description:
The Address Point Creation and verification Process: ARCBridge Consulting reviewed 15 address databases identified as source material for the MAR. Addresses of each dataset were parsed, validated, and verified against the initial Master Address Table (MAT) created using the OTR database. All unverified and invalid addresses were manually reviewed. Addresses, which were invalid due to missing address components, were further examined in the manual process. The manual review was based on the following information: Street Name, Street Range, Building Polygons, Property Scanned Images, and the Ortho. MAR addresses were also tested against the Mobile data as Mobile addresses were verified by field verification and thus more accurate. The manual review was based on the 1999 planimetric data provided by OCTO, predominantly on buildings and street centerlines. Therefore, accuracy of addresses depended somewhat on the accuracy and currentness of the planimetric data. As a result, manual review was limited by the planimetric data and the owner point X,Y coordinates provided by OCTO. The address points were primarily created from the 'Ownerpoints' database from the DC Office of Tax and Revenue (OTR) given to ARCBridge by DC OCTO. OTR: 172,470 Phase I: The 'Premiseadd' field was parsed to separate the address components. Addresses with the above mentioned components were treated as complete address. Staff checked whether the X,Y values of the complete addresses existed. If a good address didn't have X,Y values, then it was added. Addresses were manually reviewed to check the address location. Addresses, whose street number, name, and location matched, were treated as complete and added to MAR. Phase II: The addresses from the 'Address 2' field were processed, as while reviewing a number of addresses in the 'Address2' field appear correct. Staff checked whether the X,Y values of the complete addresses existed. If a good address didn't have X,Y values, then it was added. Addresses were manually reviewed to check the address location. Addresses, whose street number, name, and location matched, were treated as complete and added to MAR. Phase III: The remaining addresses that either had missing component or didn't appear correct were manually reviewed. Addresses components were then reviewed. For example: If the street type was missing or wrong and all the other components like street name, number, and location were correct, then the street type was added or changed. Staff checked whether the X,Y values of the complete addresses exist or not. If a good address didn't have X,Y values, then it was added. These modified addresses were added to the MAR. Phase IV: The remaining addresses that could not be verified, didn't appear correct, or were ambiguous were moved into the comment table. WASA: 132,868 WASA addresses didn't have X,Y values, therefore the addresses were geocoded to add X,Y coordinates. Phase I: The 'Address' field was parsed to separate the address components. Addresses with the above mentioned components were treated as complete addresses. Staff checked whether X,Y values of the complete addresses existed. If any good address didn't have XY values then it was added. Addresses were manually reviewed to check the address location. Addresses, whose street number, name & location matched, were treated as good and complete and added to MAR. Phase II. The remaining addresses that either had a missing component or didn't appear correct were manually reviewed. Addresses components were then reviewed. For example: If the street type was missing or wrong and all the other components like street name, number, and location were correct, then the street type was added or changed. Staff checked whether X,Y values of the complete addresses existed. If a good address didn't have X,Y values, then it was added. These modified addresses were then added to the MAR. Phase III. Addresses that could not be verified, didn't appear correct, or were ambiguous were moved to the comment table. MPD: 269,350 Phase I: The 'Address' field was parsed to separate the address components. Addresses with the above mentioned components were treated as complete addresses. Staff checked whether the X,Y values of the complete addresses existed. If a good address didn't have XY values, then it was added. Addresses were manually reviewed to check their location. Addresses, whose street number, name, and location matched, were treated as complete and added to MAR. Phase II: The remaining Addresses that either contained a missing component or didn't appear correct were manually reviewed. Addresses components were then reviewed. For example: If the street type was missing or wrong and all the other components like street name, number, and location were correct, then the street type was added or changed. Staff checked whether the X,Y values of the complete addresses existed. If a good address didn't have X,Y values, then it was added. These modified addresses were added to the MAR. Phase III: Addresses that couldn't be verified, didn't appear correct, or were ambiguous were moved to the comment table. Numerous MPD addresses were moved to the comments table as most of them appeared theoretical, such as a full address with no associated building. OTR- Mobile Video data: 156,527 Mobile addresses were created and verified at the physical location of the address, thus strengthening the validity of the attribution. Most other source information was generated and verified using manual in-house procedures. Mobile Video addresses were manually reviewed to check the address locations. Addresses, whose street number, name, and location matched, were treated as valid. The remaining addresses were moved to the comments table. Addresses with comments were manually reviewed. Addresses components were reviewed, for example: If type was missing or wrong and all the other components like street name, number & location were correct then type was added/changed. These modified correct addresses were moved out of the Comment table and added to the valid Mobile Video address table. Many of the same procedures were run on the remaining source datasets: DMV - Registered Vehicles - 53,507 DMV - Drivers License: - 140,595 D&B - Business: - 38,625 MCC - EMA - 195,857 MPD - Street File - 128,699 OTR- Property Owners - 129,800 OTR- Property Inventory - 161,408 BOEE- Registered Voters - 318,318 DOH - 201,796 FEMS - E 911 - 142,407 DDOT- Street Centerline - 33,291 DCRA - New Addresses - 151 Total Address Records Reviewed - 2,275,669 Working with Unverified Addresses: Phase I: The full address field of each dataset was parsed to separate the address components using the MDS Tool. Addresses with the all the required components were treated as complete addresses. All addresses validated and verified. All unverified addresses were manually reviewed to create address points. Unverified addresses were geocoded if the dataset did not have X, Y coordinates. Addresses, whose street number, name, and location matched, were treated as complete and added to MAR. Good addresses that the analyst was unable to verify were placed in the comments table. Phase II: Addresses that either had missing components or didn't appear correct were manually reviewed. Addresses components were then reviewed. For example, if the streey type was missing or incorrect and all other components, like street name, number, and location, were correct, then the streey type of the address location type was added or changed. If the street number was missing or incorrect, then the street number range and adjacent building numbers were used to correct or add the street number. If street name was missing or incorrect, then the street name at the address point location was identified and added or modified. If the quad was missing or incorrect, then the quad at the address point location was identified and added or modified. If any address was complete but the address point was at the wrong location, then staff identified the related street name and range from the street centerlines and moved theaddress point to the correct location manually. Then, the moved address points' X,Y values were updated. These modified correct addresses were added to the MAR. Phase III. Staff then completed the comments table. Addresses that couldn't be verified, didn't appear correct, or were ambiguous were moved to the comment table. It contains all of the reject addresses from the 15 source datasets. For more information regarding the construction of the MAR, refer to the "DC Addressing Final Report."