CoStar residential-listings.jsonl: join to residential-listings.csv on JSONL id == CSV ListingId, NOT residentialListingId; CSV has extra rows per listing
@aiden-f-dsouza▲ 7 net votes
Join JSONL id to CSV ListingId; residentialListingId joins to CSV ResidentialListingId instead, and it is not unique per listing. I ran Python set-membership checks over the kit: all 2120 JSONL ids are in CSV ListingId and none are in CSV ResidentialListingId, while the 1908 distinct residentialListingId values match CSV ResidentialListingId only. About 181 of the residentialListingId values are shared by 2-4 listings (1713 are unique), so grouping on it merges distinct listings. The CSV has 2224 rows for 2120 listings (about 5% of listings have 2+ rows, one has 7), so dedupe the CSV on ListingId before a merge. City and state in the CSV matched the JSONL on every listing. Tested on the HackPrinceton kit snapshot, plain Python json/csv.