[kPsarakis]

Stack: 2 of 3. #106 → this PR → #109 (metric input validation and docs). Based on #106, so the diff shows only this change. Merge #106 first.

Problem

one_to_one_hungarian() and one_to_one_greedy() return early when the results have fewer than two distinct scores, because the median threshold can't be computed then. The early return also skips the 1:1 selection itself, so:

• Tied results stay many-to-many. JaccardDistanceMatcher on two tables whose columns hold identical values scores all 4 pairs 1.0; "1:1" keeps all 4, and Precision() reports 0.5 where a 1:1 selection gives 0.0 or 1.0.
• An explicit threshold is ignored. one_to_one_hungarian(threshold=0.9) on a single 0.2-scored pair still returns that pair.

Since these are the default filters behind Precision, Recall, F1Score and PrecisionTopNPercent, those metrics are affected.

Change

• The default threshold moves into MatcherResults._default_threshold(). When all scores are equal it returns that score, so nothing is cut; otherwise it uses the same formula as before.
• Both selectors always run the 1:1 selection and always apply an explicit threshold.
• one_to_one_greedy() returns early only for empty results.
• Results with two or more distinct scores are unaffected.

Tests

New TestOneToOneTiedScores in tests/test_coverage_gaps.py, replacing TestGreedyEarlyReturn:

• tied scores stay one-to-one (greedy and Hungarian)
• an explicit threshold is applied with one distinct score
• Precision uses the 1:1 selection
• end-to-end check with JaccardDistanceMatcher

6 of the new tests fail on #106 without this change. Full suite: 318 passed locally.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 95.27%. Comparing base (957414b) to head (99f2d20).

Additional details and impacted files

@@                        Coverage Diff                         @@
##           fix/105-metric-name-collisions     #107      +/-   ##
==================================================================
+ Coverage                           95.24%   95.27%   +0.02%     
==================================================================
  Files                                  53       53              
  Lines                                2839     2833       -6     
  Branches                              446      443       -3     
==================================================================
- Hits                                 2704     2699       -5     
  Misses                                 83       83              
+ Partials                               52       51       -1     

Flag	Coverage Δ
unit	95.27% <100.00%> (+0.02%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
valentine/algorithms/matcher_results.py	98.70% <100.00%> (+0.57%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.