Our Data
An open evaluation of name-to-gender inference
We evaluate Genderize against public datasets in which each person's gender is already recorded, reporting accuracy and coverage for every set. Each individual prediction — correct, incorrect, or no prediction — is listed below, so the aggregate figures can be checked and reproduced.
Results
Per-dataset accuracy, and every prediction behind it
Every prediction sits next to the person's real gender, so you can check the numbers yourself.
Public datasets with a recorded gender for every person, scored with the person's country known.
More datasets added over time.
Showing 88,151–88,153 of 88,153
| Name | Country | True gender | Predicted | Probability | Result |
|---|---|---|---|---|---|
|
zlem Kaya
|
Türkiye
|
female
|
female
|
92.0%
|
|
|
zzet Safer
|
Türkiye
|
male
|
male
|
99.8%
|
|
|
zzet nce
|
Türkiye
|
male
|
male
|
99.8%
|
|
Method
How we measured this
Population. We score the "alive-today" set: people with a known birth year, at most 80 years old.
Accuracy is the share of returned predictions that matched the known gender. Coverage is the share of people we returned any prediction for — we abstain when there's too little signal, and those show as "no prediction".
Country context. Every prediction here is made with the person's country known — the mode most API users run, and the most accurate one.
Rounding. Figures are the exact measured value to two decimals — we never round accuracy up.