How should I tune --threshold in fuzzy_match_students.py for reliable matches?

[toughdave]

I am running fuzzy_match_students.py and want practical guidance for setting --threshold.

What is the valid range, and what threshold should I use for strict vs flexible matching?

If possible, please include a command example and links to the relevant code lines.

[toughdave]

Use --threshold as a similarity cutoff between 0.0 and 1.0.

Practical starting points

• 0.90 to 0.95: strict matching (fewer false positives)
• 0.84 to 0.89: balanced matching
• 0.75 to 0.83: flexible matching (review outputs carefully)

Example command

python3 scripts/python/reconciliation/fuzzy_match_students.py \
  --source data/sample/student_records_source.csv \
  --target data/sample/student_records_target.csv \
  --output reports/fuzzy_matches.csv \
  --summary reports/fuzzy_matches_summary.json \
  --threshold 0.86

Relevant code references

• CLI threshold definition: fuzzy_match_students.py#L151-L156
• Range validation in main(): fuzzy_match_students.py#L163-L164