app.py

import gradio as gr
import pandas as pd
from jamo import h2j, j2hcj
import dill as pickle

enc = pickle.load(open('encoder.pkl','rb'))
model = pickle.load(open('kornamegender.model','rb'))

def process(name):
    name = list(name)
    name = name + [None] * (3 - len(name))

    p1, p2, p3 = list(), list(), list()

    p1 = list(j2hcj(h2j(name[0])))
    p1 = p1 + [None] * (3 - len(p1))    

    if name[1] == None:
        p2 = [None, None, None]
    else:
        p2 = list(j2hcj(h2j(name[1])))
        p2 = p2 + [None] * (3 - len(p2))

    if name[2] == None:
        p3 = [None, None, None]
    else:
        p3 = list(j2hcj(h2j(name[2])))
        p3 = p3 + [None] * (3 - len(p3))

    p = p1 + p2 + p3

    p = pd.DataFrame(p).T

    en = enc.transform(p)

    holder = []

    if model.predict(en) == 0:
        holder.append('Male')
    else:
        holder.append('Female')
    holder.append("{0:.3f}%".format(max(model.predict_proba(en)[0])*100))

    return (holder[0], holder[1])

with gr.Blocks() as namegender:
    
    gr.Markdown("<h3>Welcome to David's KDI School Machine Learning Project!</h3><h1>Predicting Gender Using <u>Hanja-based</u> Korean Names</h1>")
    gr.Markdown("Type in a name (following the Acceptable Name Criteria) and see what the gender is predicted to be!")
    
    with gr.Row():
        with gr.Column():
            with gr.Row():
                input=gr.Textbox(label="Korean Name", placeholder = "Only Korean Names (Ex: 형관)")
                button = gr.Button("Go!")
            with gr.Row():            
                xgbmodel=[gr.Textbox(label = "Gender Prediction"), gr.Textbox(label = "Confidence")]
        with gr.Column():
            gr.Markdown("<h3>Acceptable Name Criteria:</h3>" + 
            "- Typed Strictly in Korean <strong>and</strong> Hanja-based<br />"+
            "- Must avoid non-valid name characters (ex. 쀏)<br />"+
            "- Must not contain the surname<br />"+
            "- Must be 3 characters or fewer<br /><br />")
    input.submit(process, inputs=input, outputs = xgbmodel)
    button.click(process, inputs=input, outputs = xgbmodel)
    
    gr.Markdown("<strong>Valid</strong> Examples: 정심, 재혁, 소영, 경진, 화준, 설, 뜰에봄<br/><strong>Invalid</strong> Examples: 뤫ㅇ, 이자벨라, 그라미, 홍길동, David")
    gr.Markdown("<h3>Other notes</h3>"+
                "<i>Non-Hanja and other invalid names will either result in an error or an invalid result.<br />"+
                "The closer 'Confidence' is to 50%, the more gender-neutral the name becomes.<br />"+
                "This model assumes only the two traditional genders of 'male' and 'female'.</i>")
    
namegender.launch(share=True)