For testing, compare the actual PSGC vs. the expected PSGC
Reduce the threshold to 85% temporarily
Increase the time for the messy datasets to 180s (based on the current ngrams matcher run)
The current matcher is 88% accurate on our messy dataset
Creating test database for alias 'default'...
System check identified no issues (1 silenced).
Testing All Matches...
Testing Missing Barangay...
Testing Missing MuniCity...
Testing Missing Province...
Testing Just Barangay...
Testing Just MuniCity...
Testing Just Province...
Testing Barangay-MuniCity...
Testing MuniCity-Province...
Testing Barangay-Province...
Testing Naming Variations...
Testing No matches...
.Found 2000 correct matches out of 2000 records
Clean:
Duration: 7.62
Accuracy: 1.00
..Found 1759 correct matches out of 2000 records
Messy:
Duration: 129.65
Accuracy: 0.88
...
----------------------------------------------------------------------
Ran 6 tests in 175.478s
OK
Destroying test database for alias 'default'...