🌍 Name → Country
A name-to-country classifier, presented as a walkthrough of its ML pipeline — a statistical guess from character patterns, not a fact about any person.
A guided tour of the pipeline
This Space walks through the whole project, one page per stage — the same stages that
live in pipeline/:
| # | Page | Script | What it does |
|---|---|---|---|
| 1 | EDA | eda01.py |
Profile 33M raw rows + the generated splits; quantify class balance and name ambiguity |
| 2 | Preprocess | preprocess02.py |
Synthesize full names, split 70/15/15, featurize to index arrays |
| 3 | Baselines | baseline03.py |
Train 3 sklearn TF-IDF models as a CPU floor |
| 4 | Evaluation | evaluation04.py |
Score on the held-out test set (strict + ambiguity-aware) |
| 5 | Predict | predict05.py |
Live inference — try your own names |
On each stage page, hit ▶ Run this step to replay that stage's console log, then see the pre-generated plots and metrics it produced. (The heavy stages ran offline over millions of rows — only Predict runs live here.)
📦 Source code: github.com/britoddd/ml-name_country_origin_checker
⚠️ Statistical guess only. This predicts likely countries from the character patterns of a romanized name — it is not a statement about any real person's nationality, ethnicity, or origin. Many names span multiple countries. Do not use for profiling, hiring, credit, immigration, or law-enforcement decisions. Names you enter are not logged or stored.