backend/data_generator.py

import pandas as pd
import numpy as np
import random
import os

# Configuration for Moroccan Real Estate High-Fidelity (Mubawab/ANCFCC/BAM 2024)
CITIES = {
    "Casablanca": {
        "base_price_apt": 19500, # Increased based on 2024 trends
        "base_price_villa": 27500,
        "premium_neighborhoods": ["Anfa", "Ain Diab", "Gauthier", "Les Princesses", "Racine", "California"],
        "standard_neighborhoods": ["Maarif", "Sidi Maârouf", "Oulfa", "Bernoussi", "Belvédère", "Salmia"],
        "has_tram": True
    },
    "Rabat": {
        "base_price_apt": 17800, # Rabat prices are high due to administrative status
        "base_price_villa": 24000,
        "premium_neighborhoods": ["Hay Riad", "Souissi", "Agdal", "Ambassadeurs", "Orangers"],
        "standard_neighborhoods": ["Hassan", "Yacoub El Mansour", "El Menzeh", "Ocean", "Kamra"],
        "has_tram": True
    },
    "Marrakech": {
        "base_price_apt": 14500, 
        "base_price_villa": 25000,
        "premium_neighborhoods": ["Hivernage", "Palmeraie", "Gueliz-High", "Amelkis", "Targa-Premium"],
        "standard_neighborhoods": ["Gueliz-Standard", "Medina", "Targa", "Mhamid", "Massira", "Iziki"],
        "has_tram": False
    },
    "Tanger": {
        "base_price_apt": 15500, 
        "base_price_villa": 19000,
        "premium_neighborhoods": ["Malabata", "Marshane", "Achakkar", "Jebel Kebir", "California-Tanger"],
        "standard_neighborhoods": ["Iberia", "Val Fleuri", "Beni Makada", "Dradeb", "Moghogha"],
        "has_tram": False
    },
    "Agadir": {
        "base_price_apt": 16500,
        "base_price_villa": 20000,
        "premium_neighborhoods": ["Founty", "Charaf", "Haut Founty", "Marina"],
        "standard_neighborhoods": ["Dakhla", "Salam", "Anza", "Al Houda"],
        "has_tram": False
    }
}

PROPERTY_TYPES = ["Appartement", "Villa", "Maison"]
STANDINGS = ["Haut Standing", "Moyen Standing", "Economique"]
CONDITIONS = ["Neuf", "Bon état", "A rénover"]
ORIENTATIONS = ["Sud (Ensoleillé)", "Est", "Ouest", "Nord"]
VIEWS = ["Sans vis-à-vis", "Vue sur mer", "Vue sur Parc/Jardin", "Vue sur rue"]
RESIDENCY_TYPES = ["Résidence fermée & sécurisée", "Public / Quartier ouvert"]

def generate_data(n_samples=20000):
    data = []
    
    for _ in range(n_samples):
        city = random.choice(list(CITIES.keys()))
        is_premium_zone = random.random() < 0.35
        neighborhoods = CITIES[city]["premium_neighborhoods"] if is_premium_zone else CITIES[city]["standard_neighborhoods"]
        neighborhood = random.choice(neighborhoods)
        
        prop_type = random.choice(PROPERTY_TYPES)
        standing = random.choice(STANDINGS)
        condition = random.choice(CONDITIONS)
        orientation = random.choice(ORIENTATIONS)
        view = random.choice(VIEWS)
        residency = random.choice(RESIDENCY_TYPES)
        
        # Dimensions logic
        if prop_type == "Villa":
            surface = random.randint(200, 1500)
            rooms = random.randint(5, 15)
            bedrooms = random.randint(3, 8)
            floor = 0
            base_price_m2 = CITIES[city]["base_price_villa"]
        else:
            # Apartment specific
            surface = random.randint(40, 350)
            rooms = random.randint(1, 8)
            bedrooms = random.randint(1, 5)
            floor = random.randint(0, 12)
            base_price_m2 = CITIES[city]["base_price_apt"]
            
        # Specific Moroccan features
        has_lift = 1 if (prop_type == "Appartement" and floor > 2) or (is_premium_zone and prop_type == "Appartement") else 0
        has_pool = 1 if (prop_type == "Villa" and (is_premium_zone or random.random() > 0.5)) else 0
        has_garden = 1 if (prop_type == "Villa" or (prop_type == "Appartement" and floor == 0 and random.random() > 0.7)) else 0
        parking_spots = random.randint(1, 3) if (is_premium_zone or standing == "Haut Standing") else random.randint(0, 1)
        
        # New Proximity Logic
        proximity_tram = 1 if (CITIES[city]["has_tram"] and random.random() > 0.6) else 0
        proximity_university = 1 if (random.random() > 0.7) else 0 # High demand for students
        proximity_mosque = 1 if (random.random() > 0.3) else 0 # Essential service in Moroccan neighborhoods
        
        # --- Valuation Logic (High-Fidelity) ---
        # Multipliers
        mult = 1.0
        
        # Neighborhood & Standing Impact
        if is_premium_zone: mult *= 1.6
        if neighborhood in ["Souissi", "Anfa", "Hivernage", "Ain Diab", "Souissi"]: mult *= 1.4
        
        # Standing (Standard finishes vs Luxury)
        standing_map = {"Haut Standing": 1.5, "Moyen Standing": 1.0, "Economique": 0.6}
        mult *= standing_map[standing]
        
        # Residency & Security
        if residency == "Résidence fermée & sécurisée": mult *= 1.15
        
        # Orientation (Sud is highly valued in MOROCCO for winter sun)
        if orientation == "Sud (Ensoleillé)": mult *= 1.08
        elif orientation == "Nord": mult *= 0.95
        
        # View
        view_map = {"Vue sur mer": 1.35, "Vue sur Parc/Jardin": 1.12, "Sans vis-à-vis": 1.10, "Vue sur rue": 0.95}
        mult *= view_map[view]
        
        # Condition
        cond_map = {"Neuf": 1.25, "Bon état": 1.0, "A rénover": 0.7}
        mult *= cond_map[condition]
        
        # Proximity Impact
        if proximity_tram == 1: mult *= 1.05
        if proximity_university == 1: mult *= 1.07 # Student demand premium
        if proximity_mosque == 1: mult *= 1.04 # Convenience premium
        
        # Extras
        if has_pool: mult *= 1.2
        if has_garden: mult *= 1.1
        
        # Final calculation
        price_per_m2 = base_price_m2 * mult
        total_price = (price_per_m2 * surface) * random.uniform(0.97, 1.03)
        
        data.append({
            "City": city,
            "Neighborhood": neighborhood,
            "Type": prop_type,
            "Surface": surface,
            "Rooms": rooms,
            "Bedrooms": bedrooms,
            "Standing": standing,
            "Residency": residency,
            "Orientation": orientation,
            "View": view,
            "Condition": condition,
            "Floor": floor,
            "Lift": int(has_lift),
            "Pool": int(has_pool),
            "Garden": int(has_garden),
            "Parking_Spots": parking_spots,
            "Proximity_Tram": int(proximity_tram),
            "Proximity_University": int(proximity_university),
            "Proximity_Mosque": int(proximity_mosque),
            "Price": round(total_price, -3)
        })
        
    df = pd.DataFrame(data)
    output_path = os.path.join(os.path.dirname(__file__), "..", "data", "morocco_real_estate_data_pro.csv")
    df.to_csv(output_path, index=False, encoding='utf-8-sig')
    print(f"Generated {n_samples} High-Fidelity records in {output_path}")

if __name__ == "__main__":
    generate_data(20000)