CoolFace
Modelpublic

hongducle96/vietnamese-name-gender-classification

sourceHugging Faceapache-2.0updated 10mo agoView on Hugging Face
0likes
Model Card

Vietnamese Name Gender Classification

Model name: vn-gender-name-classification Model type: Classical ML (TF-IDF + Logistic Regression + Linear SVM) Task: Gender classification from Vietnamese full names Author: Le Hong Duc - hongducle96@gmail.com - https://www.linkedin.com/in/hongduc96/

1. Overview

This model predicts binary gender (Nam / Nữ – male/female) from Vietnamese full names. It is trained on a dataset of Vietnamese names labeled with binary gender, using character n-gram TF-IDF features and two base models:

  • —Logistic Regression (LR)
  • —Linear SVM This model is designed to be:
  • —Lightweight – classical ML, no GPUs required
  • —Easy to integrate – scikit-learn pipelines saved as .joblib
  • —Practical – includes a clear production decision rule and an ambiguity flag
  • —Accuracy Model: ~96.5%

2. Training data

  • —Source: collection of Vietnamese full names with gender labels on a public website
  • —Labels: Giới tính ∈ {Nam, Nữ} (binary male/female)
  • —Size:
  • —Training: ~50,000 labeled names
  • —Test: ~15,000 names (held-out evaluation set)

2.1. Preprocessing

For each name:

  • —Normalize whitespace (strip, collapse multiple spaces).
  • —Lowercase the text, but keep Vietnamese accents (diacritics).
  • —Create an additional variant name_wo_surname by:
  • —Splitting on spaces.
  • —Dropping the first token (surname).
  • —Keeping middle + given name. Example:
  • —"Nguyễn Hoàng Phúc" →
  • —full_name_clean = "nguyễn hoàng phúc"
  • —name_wo_surname = "hoàng phúc" Both forms are used for training to make the model more robust to how names are provided.

2.2. Evaluation

On the held-out test set (~15K names), with results:

  • —Accuracy: ~96.5%
  • —Proportion of names marked ambiguous: ~2.6% | Class | Precision | Recall | F1-score | Support | |-------------|-----------|--------|----------|---------| | Nam | 0.956 | 0.964 | 0.960 | 6637 | | Nữ | 0.972 | 0.965 | 0.968 | 8409 | | Accuracy| | | 0.965| 15046 | | Macro avg | 0.964 | 0.965 | 0.964 | 15046 | | Weighted avg | 0.965 | 0.965 | 0.965 | 15046 |

Confusion matrix (Nam/Nữ) – PROD:

Actual \ PredictedNamNữ
Nam6401236
Nữ2988111
  • —Nam: 6,401 correctly predicted vs. 236 misclassified as “Nữ”.
  • —Nữ: 8,111 correctly predicted vs. 298 misclassified as “Nam”. Number of names marked ambiguous by PROD: 395 / 15046 (2.63%)

3. Model architecture

3.1. Features

Both LR and SVM use the same character-level TF-IDF features:

  • —TfidfVectorizer with:
  • —analyzer="char_wb" (character n-grams within word boundaries)
  • —ngram_range=(2, 6) (bigrams to 6-grams)
  • —min_df=2 (ignore n-grams that appear in fewer than 2 documents) This allows the model to learn patterns from substrings of names, which is very effective for Vietnamese given names.

3.2. Base models

Two scikit-learn pipelines are trained:

  • —model_lr.joblib
  • —TfidfVectorizer → LogisticRegression(max_iter=1000, n_jobs=-1)
  • —Supports predict_proba for probability estimates.
  • —model_svm.joblib
  • —TfidfVectorizer → LinearSVC
  • —Strong linear classifier, but outputs only class labels (no probabilities).

3.3. Production ensemble logic (simplified)

For an input full name full_name_raw:

  1. 1.Preprocessing
  2. 2.full_name_clean = normalize_name_keep_accent(full_name_raw)
  3. 3.name_wo_surname = drop_surname(full_name_clean)
  4. 4.SVM predictions
  5. 5.SVM_FULL = model_svm.predict([full_name_clean])
  6. 6.SVM_WO = model_svm.predict([name_wo_surname])
  7. 7.LR combined probabilities
  8. 8.Get probabilities from LR on both text variants:
  9. 9.p_Nam_full, p_Nu_full = LR(full_name_clean)
  10. 10.p_Nam_wo, p_Nu_wo = LR(name_wo_surname)
  11. 11.Average the Nam probability:
  12. 12.p_Nam_avg = 0.5 * (p_Nam_full + p_Nam_wo)
  13. 13.p_Nu_avg = 1.0 - p_Nam_avg
  14. 14.LR_label = "Nam" if p_Nam_avg >= p_Nu_avg else "Nữ"
  15. 15.LR_conf = max(p_Nam_avg, p_Nu_avg)
  1. 1.Decision rule
  • —If SVM_FULL == SVM_WO (SVMs agree):
  • —final_label = SVM_FULL
  • —Mark as ambiguous if LR_conf < lr_conf_clear (default 0.60).
  • —If SVM_FULL != SVM_WO (SVMs disagree, hard case):
  • —If LR_conf >= lr_conf_as_referee (default 0.70), follow LR: final_label = LR_label
  • —Otherwise, default to SVM_WO: final_label = SVM_WO
  • —Mark as ambiguous = True. The final output includes:
  • —final_label – production gender prediction ("Nam" or "Nữ")
  • —LR_conf – LR combined confidence
  • —is_ambiguous – whether this name is considered ambiguous by the ensemble

Intended use & limitations Intended use

  • —Analytics and statistical exploration of Vietnamese name data.
  • —Educational demos of classical ML methods on Vietnamese text.
  • —Lightweight baseline for experiments involving Vietnamese full names.

Limitations

  • —The model predicts only binary gender (Nam / Nữ) from naming patterns.
  • —It does not capture the full diversity of gender identities or self-determined gender.
  • —It may be inaccurate or biased for:
  • —Rare or gender-neutral names
  • —Non-Vietnamese names, abbreviations, nicknames
  • —Names from under-represented regions or time periods in the training data

This model should not be used for high-stakes decisions that affect individual rights, access to services, or opportunities, or in any context where misclassification could cause harm or discrimination.