owuraku427/easyvisa-xgb

EasyVisa XGBoost Model

Tuned XGBoost classifier for US visa application approval prediction (Certified vs Denied).

Live demo: owuraku427/easyvisa-demo

Files

• easyvisa_xgb.joblib — serialized scikit-learn-compatible XGBoost model
• schema.json — feature column order and category definitions required for inference

Usage

import joblib, json, pandas as pd
from huggingface_hub import hf_hub_download

model_path = hf_hub_download("owuraku427/easyvisa-xgb", "easyvisa_xgb.joblib")
schema_path = hf_hub_download("owuraku427/easyvisa-xgb", "schema.json")

model = joblib.load(model_path)
schema = json.load(open(schema_path))

# Build a row matching schema["feature_columns"], then:
# model.predict(row)

Performance

• F1 (test set): ~0.83
• Trained on 25,480 historical applications

Limitations

Trained on FY 2016 data. Not legal advice — for screening only.