bcgov / bcgov/ols-geocoder

Missing addresses not found in rejected list

Open
#371 2 comments 0 reactions 0 assignees View on GitHub
bug data integration geocoder
Dominant language
Java
Stars
17
Forks
7
Avg merge
10h 2m
Merged PRs (30d)
15

Description

**Scenario:**
*Geocoded in PROD with October 2023 data vintage.

Searching for the following addressString ([50 Clark Dr, Port Alice, BC](https://geocoder.api.gov.bc.ca/addresses.html?addressString=50%20Clark%20Dr,%20Port%20Alice,%20BC)) results in a BLOCK level match with a fault of STREET_NAME.spelledWrong. The resulting fullAddress includes an updated street name of Clark**e** Dr.

**Task:**
Investigate why the addresses on Clark Dr were not found in the rejected address SHPs.

**Initial Review:**

It was found that BCA, AddressBC and the street sign shown in Google Street View use the street name ‘Clark’. [BCA's assessment search tool](https://www.bcassessment.ca/Property/AssessmentSearch?) also lists ’50 CLARK DR PORT ALICE’ with a PID. The PID was also confirmed in the PMBC and BCA input files. This street name discrepancy has been submitted to the road network team for review.

Based on the street name difference, it was no surprise that 50 Clark Dr, Port Alice, BC was not listed in the site_Hybrid_geocoder file. However, it was also not listed in the rejected address SHPs.

Next, the PMBC layer was examined and showed that this location is a strata lot. More specifically, there were 102 parcel polygons stacked at this location. 101 parcels had a unique PID and an owner type of Private. One parcel was unclassified and had no PID.

Performing a within search (in PROD) of this parcel revealed 69 sites at -127.4830565, 50.4290903. Based on the site_Hybrid_geocoder file they each have a unique PID.

Regarding the address range, most addresses appear to be within the correct overall range. However, the road network shows that part of Clark**e** Dr only has addresses on the left side of the street (right side range is 0 - 0), while having addresses on both sides of the street in another section. This corresponds with the residential driveways seen in aerial photos and Google Street View.

Finally, as a test the source addresses from AddressBC found within the associated parcel were Batch Geocoded. Of the 99 addresses, 100% of the non-Clark Dr addresses (67x) were found with a matchPrecision of CIVIC_NUMBER. The Clark Dr addresses only matched to the BLOCK level (30x) and the STREET level (2x - CIVIC_NUMBER.notInAnyBlock).

Contributor guide

No contributing guide indexed for this repository

Research direction

Start by comparing the PMBC layer and AddressBC source addresses for the Clark Dr parcel with the site_Hybrid_geocoder file and rejected address SHPs. Reproduce the PROD batch-geocoding results and trace how the 99 source addresses are classified. Done means explaining why the Clark Dr addresses were omitted from the rejected outputs and identifying the affected processing stage.

Written by the indexing model from the issue text.

Assessment

Tech stack
java
Domain
data-engineering, search
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.