app.py

import gradio as gr
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression

# ------------------------
# 1. 데이터 로드
# ------------------------
df = pd.read_excel("Restaurants.xlsx")
df["가격대"] = df["가격대"].astype(float)

# Synthetic label 생성
ideal_budget = 12000
df["budget_diff"] = abs(df["가격대"] - ideal_budget)

df["label"] = np.where(
    (df["평점"] >= 4.0) & (df["budget_diff"] <= 3000),
    1,
    0
)

# 머신러닝용 인코딩
df_ml = pd.get_dummies(df, columns=["음식종류", "연령층", "방문목적"])

X = df_ml.drop(columns=["식당명", "리뷰", "위치(지하철역)", "label"])
y = df_ml["label"]

X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42
)

# ------------------------
# 2. ML 모델 (HuggingFace에서 100% 실행 가능한 LogisticRegression)
# ------------------------
model = LogisticRegression(max_iter=500)
model.fit(X_train, y_train)

df["predict_score"] = model.predict_proba(X)[:, 1]


# ------------------------
# 3. 추천 함수
# ------------------------
def recommend_ai(region, food_type, budget, age):

    # ----- 기본 필터 -----
    filtered = df[(df["위치(지하철역)"] == region) &
                  (df["음식종류"] == food_type)]

    if filtered.empty:
        return "조건에 맞는 식당이 없습니다."

    filtered = filtered.copy()

    # ----- 🔥 예산 필터: budget - 3000 ~ budget + 3000 -----
    min_price = budget - 3000
    max_price = budget + 3000

    filtered = filtered[(filtered["가격대"] >= min_price) &
                        (filtered["가격대"] <= max_price)]

    if filtered.empty:
        return f"예산 범위({min_price}원 ~ {max_price}원)에 맞는 식당이 없습니다."

    # 예산 차이
    filtered["user_budget_diff"] = abs(filtered["가격대"] - budget)

    # 최종 점수 계산
    filtered["final_score"] = (
        filtered["predict_score"] * 0.7 +
        (1 / (filtered["user_budget_diff"] + 1)) * 0.3
    )

    # Top 5
    result = filtered.sort_values("final_score", ascending=False).head(5)
    top = result.iloc[0]

    # -----------------------------
    # 🏆 TOP 1 출력
    # -----------------------------
    output = ""
    output += "<div style='font-size:20px; font-weight:bold; margin-bottom:10px;'>"
    output += "🏆✨ <b>AI의 1등 추천 식당!</b> ✨🏆"
    output += "</div>"

    output += "<div style='font-size:16px; line-height:1.7;'>"
    output += f"⭐ <b>{top['식당명']}</b><br>"
    output += f"📍 위치: {top['위치(지하철역)']}<br>"
    output += f"🍱 종류: {top['음식종류']}<br>"
    output += f"💰 가격대: {int(top['가격대'])}원<br>"
    output += f"⭐ 평점: {top['평점']}<br>"
    output += f"💬 리뷰: {top['리뷰']}<br>"
    output += "</div>"

    # ----- 텍스트 기반 구분선 -----
    output += "<div style='font-family:monospace; margin: 12px 0 10px 0;'>"
    output += "=" * 80
    output += "</div>"

    # -----------------------------
    # 📌 그 외 추천 식당들
    # -----------------------------
    output += "<div style='margin-top:5px; margin-bottom:10px;'>"
    output += "<h3 style='margin:0; padding:0;'>📌 그 외 추천 식당들</h3>"
    output += "</div>"

    for _, row in result.iloc[1:].iterrows():
        output += "<div style='font-size:14px; line-height:1.55; margin-top:8px; margin-bottom:12px;'>"
        output += f"⭐ {row['식당명']}<br>"
        output += f"🍱 {row['음식종류']} | 💰 {int(row['가격대'])}원 | ⭐ {row['평점']}<br>"
        output += f"💬 {row['리뷰']}<br>"
        output += "<div style='margin:10px 0 6px 0; border-bottom:1px solid #ccc;'></div>"
        output += "</div>"

    return output


# ------------------------
# 4. Gradio UI
# ------------------------
with gr.Blocks() as demo:
    gr.Markdown("## 🤖 AI 기반 머신러닝 서울 맛집 추천 시스템")

    region = gr.Dropdown(
        choices=sorted(df["위치(지하철역)"].unique()),
        label="지하철역 선택"
    )
    food_type = gr.Dropdown(
        choices=sorted(df["음식종류"].unique()),
        label="음식 종류"
    )
    budget = gr.Slider(
        5000, 30000, value=12000, step=500,
        label="예산(원): 오차범위 ± 3,000원을 적용합니다."
    )
    age = gr.Dropdown(
        choices=sorted(df["연령층"].unique()),
        label="연령층"
    )

    btn = gr.Button("🔍 AI 추천받기")
    output_box = gr.HTML()

    btn.click(
        recommend_ai,
        inputs=[region, food_type, budget, age],
        outputs=output_box
    )