Changelog • fedmatch

Fixed documentation referencing “X” to be to the proper vignette
Fixed documentation referencing “wgt_jaccard_distance” that should be “wgt_jaccard_dist”
Fixed issue so that multivar_match warns when invalid compare_types are used
Fixed issue where wgt_jaccard_dist wasn’t listed as a valid compare type

Fixed bug in build_tier_settings that didn’t include sequential_words argument
Fixed bug where clean_strings wasn’t removing the replacement words even if remove_words == T
Fixed bug where tier_match didn’t work in some cases with fuzzy_match
Now exporting build_corpus to easily compute wgt_jaccard_distance
Added weighted jaccard distance option to compare_type in multivar matching

Added wgt_jaccard_distance to easily compute Weighted Jaccard Distances for two string vectors
Lowered R requirement to 3.5.3
Added behavior to remove NA observations from exact matches, reinserting them into later tiers if necessary
Fixed bug in tier_match building ‘nomatch’ datasets incorrectly
Fixed bug in wgt_jaccard where inverse frequencies were not being logged.

Fixed STATE_FIPS to actually include FIPS codes
Changed the output column from ‘multivar’ matching to be called ‘multivar_score’ rather than ‘matchscore’, so as not to conflict with score_settings in merge_plus
Fixed bug in Weighted Jaccard scoring that would allow scores > 1 if the two names shared multiples of the same word

Fixed bugs in Weighted Jaccard code to run on MacOS environments

Fixed bugs in Makevars so that C++ code runs in parallel
Fixed bugs in tier_match that stopped matches from being removed in between tiers
Fixed bugs preventing the successful build on other platforms besides native linux

Added full data.table functionality to dramatically increase speed
Added ‘multivar’ matching to match on multiple variables
- Added capability to train a logit model and use the output
Added ‘Weighted Jaccard’ match to get better matches