app.py

import os
import pandas as pd
import numpy as np
import plotly.express as px
import plotly.graph_objects as go
import folium
import gradio as gr
from folium.plugins import MarkerCluster

def generate_nursery_data(num=5500):
    np.random.seed(42)
    governorates = ["Cairo", "Alexandria", "Giza", "Aswan", "Luxor", "Dakahlia", "Sharqia", "Qalyubia",
        "Kafr El Sheikh", "Gharbia", "Menoufia", "Beheira", "Ismailia", "Suez", "Port Said", "Damietta",
        "Fayoum", "Beni Suef", "Minya", "Assiut", "Sohag", "Qena", "Red Sea", "New Valley", "Matruh",
        "North Sinai", "South Sinai"]
    df = pd.DataFrame()
    df["Governorate"] = np.random.choice(governorates, num)
    df["Licensed"] = np.random.choice(["Licensed", "Unlicensed"], num, p=[0.62, 0.38])
    df["Unlicensed_Status"] = np.where(df["Licensed"] == "Unlicensed",
        np.random.choice(["No Application", "Licensing in Progress", "License Rejected"], num), "Not Applicable")
    df["Rejection_Reason"] = np.where(df["Unlicensed_Status"] == "License Rejected",
        np.random.choice(["Missing Documents", "Non-compliant Facility", "Safety Violations", "Staff Unqualified", "Others"], num), "Not Applicable")
    df["Ownership"] = np.random.choice(["Rent", "Endowment", "Owned", "Usufruct", "Others"], num, p=[0.3,0.1,0.4,0.15,0.05])
    df["Affiliation"] = np.random.choice(["Governmental", "Private", "CBO", "Individual"], num, p=[0.25,0.35,0.2,0.2])
    df["Total_Classes"] = np.random.randint(1, 10, num)
    df["Avg_Class_Area_sqm"] = np.random.normal(25, 5, num).round(1)
    df["Avg_Children_Per_Class"] = np.random.randint(10, 35, num)
    df["Avg_Occupancy_Rate"] = np.random.uniform(0.6, 1.0, num).round(2)
    df["Total_Employees"] = np.random.randint(5, 25, num)
    df["Edu_Staff_to_Children"] = np.random.uniform(0.05, 0.2, num).round(2)
    df["Employee_Support_%"] = np.random.uniform(10, 30, num).round(1)
    df["Employee_Edu_%"] = np.random.uniform(50, 80, num).round(1)
    df["Employee_Admin_%"] = (100 - (df["Employee_Support_%"] + df["Employee_Edu_%"]).clip(upper=100)).round(1)
    for col in ["Academic_Activities", "Languages", "Motor_Sensory", "Music_Theater", "Religious_Education", "Arts_Crafts", "Other_Activities"]:
        df[col] = np.random.choice([True, False], num)
    df["Meal_Subscription"] = np.random.choice(["Yes", "No"], num, p=[0.6, 0.4])
    bins = ["0-50", "50-250", "250-500", "500-1000", "1000-2500", "2500+"]
    df["Meal_Subscription_Value"] = np.where(df["Meal_Subscription"] == "Yes",
        np.random.choice(bins, num, p=[0.05, 0.4, 0.3, 0.15, 0.08, 0.02]), "Not Applicable")
    return df

# === Full Dashboard Launcher ===
def launch_dashboard():
    df = generate_nursery_data()

    gender_fig = px.pie(names=["Male", "Female"], values=[43, 57], title="Gender Distribution")

    enrollment_fig = px.bar(
        x=["Enrollment Rate", "Licensed %", "Unlicensed %"],
        y=[10, (df['Licensed'] == 'Licensed').mean()*100, (df['Licensed'] == 'Unlicensed').mean()*100],
        title="Enrollment & Licensing Overview",
        labels={"x": "Metric", "y": "Percentage"},
        text_auto=True
    )

    counts_fig = px.bar(
        x=["Total Nurseries", "Children Enrolled"],
        y=[5500, 55000],
        title="Total Counts",
        text_auto=True
    )

    ownership_fig = px.pie(df, names="Ownership", title="Ownership Distribution")
    affiliation_fig = px.pie(df, names="Affiliation", title="Affiliation Distribution")

    activity_cols = ["Academic_Activities", "Languages", "Motor_Sensory", "Music_Theater", "Religious_Education", "Arts_Crafts", "Other_Activities"]
    activity_counts = df[activity_cols].apply(pd.Series.value_counts).T.rename(columns={True: "Offered", False: "Not Offered"})
    activity_counts["Activity"] = activity_counts.index
    activity_fig = px.bar(activity_counts, x="Activity", y="Offered", title="Activities Offered")

    meal_fig = px.histogram(df[df["Meal_Subscription"] == "Yes"], x="Meal_Subscription_Value", title="Meal Subscription Value Distribution")
    classroom_fig = px.histogram(df, x="Avg_Class_Area_sqm", nbins=30, title="Classroom Area Distribution")

    emp_stack = go.Figure()
    for role in ["Employee_Support_%", "Employee_Edu_%", "Employee_Admin_%"]:
        emp_stack.add_trace(go.Box(y=df[role], name=role))
    emp_stack.update_layout(title="Employee Role Distribution")

    gov_summary = df.groupby("Governorate").agg(
        Count=("Governorate", "count"),
        Licensed_Rate=("Licensed", lambda x: (x=="Licensed").mean()*100),
        Activity_Score=("Academic_Activities", lambda x: (x.sum() / len(x)) * 100),
        Rejection_Score=("Rejection_Reason", lambda x: (x != 'Not Applicable').mean() * 100)
    ).reset_index()

    gov_summary["Compliance_Tier"] = np.where(
        gov_summary["Licensed_Rate"] < 50, "Low",
        np.where(gov_summary["Licensed_Rate"] < 75, "Medium", "High")
    )

    gov_bar = px.bar(gov_summary, x="Governorate", y="Licensed_Rate", title="Licensing % by Governorate")
    reject_fig = px.histogram(df[df['Rejection_Reason'] != 'Not Applicable'], x="Governorate", color="Rejection_Reason", title="Rejection Reasons by Governorate")
    lang_fig = px.histogram(df[df['Languages']], x="Governorate", title="Language Activities Offered")
    music_fig = px.histogram(df[df['Music_Theater']], x="Governorate", title="Music & Theater Offered")
    tier_fig = px.bar(gov_summary, x="Governorate", y="Licensed_Rate", color="Compliance_Tier", title="Compliance Score by Governorate")

    gov_coords = {g: [26.8 + np.random.rand()/2, 30.8 + np.random.rand()/2] for g in df["Governorate"].unique()}
    maps = {}
    for label, col, colors in [
        ("Licensing Coverage", "Licensed", {"Licensed": "green", "Unlicensed": "red"}),
        ("Ownership Type", "Ownership", {}),
        ("Affiliation Type", "Affiliation", {})
    ]:
        fmap = folium.Map(location=[26.8, 30.8], zoom_start=6, control_scale=True, tiles='CartoDB positron')
        mcluster = MarkerCluster().add_to(fmap)
        for _, row in df.iterrows():
            gov = row['Governorate']
            coords = gov_coords.get(gov, [26.8, 30.8])
            value = row[col]
            color = colors.get(value, "blue")
            popup = folium.Popup(f"<b>{gov}</b><br>{label}: {value}", max_width=300)
            folium.Marker(location=coords, popup=popup, icon=folium.Icon(color=color)).add_to(mcluster)
        file = f"map_{col}.html"
        fmap.save(file)
        maps[label] = file

    insights = """
### Insights & Strategic Interventions
- **Licensing Focus**: Target Cairo, Giza, and Sharqia with <50% licensed nurseries
- **Facility Improvement**: Prioritize Governorates with smaller average classroom area (<22 sqm)
- **Rejection Reduction**: Most frequent reasons: 'Safety Violations' and 'Staff Unqualified'
- **Activity Gaps**: Arts, Music, and Sensory activities need boosting in rural areas
- **Staffing Ratios**: Recommend raising educational staff-to-child ratios above 0.12 minimum

### Priority Interventions by Tier:
- **Urgent** (Low Compliance): Cairo, Qalyubia — fast-track licensing, safety training
- **Strategic** (Medium Compliance): Beheira, Minya — curriculum and activity upgrade
- **Maintain** (High Compliance): Luxor, Red Sea — monitoring and recognition

### Expected Impact:
- Increase licensing compliance by 20% in 6 months
- Improve readiness of 1,500 nurseries via safety/teacher interventions
- Boost developmental outcomes by expanding educational offerings
"""

    with gr.Blocks() as demo:
        gr.Markdown("# 🇪🇬 Egypt Nursery Dashboard – July 2025")
        charts = [counts_fig, enrollment_fig, gender_fig, ownership_fig, affiliation_fig, activity_fig,
                  meal_fig, classroom_fig, emp_stack, gov_bar, reject_fig, lang_fig, music_fig, tier_fig]
        titles = ["Total Counts", "Enrollment & Licensing Overview", "Gender Distribution", "Ownership", "Affiliation",
                  "Activities Offered", "Meal Pricing", "Classroom Area", "Employee Roles", "Licensing by Governorate",
                  "Rejection Reasons", "Language Activities", "Music & Theater", "Compliance Tier"]
        for title, fig in zip(titles, charts):
            gr.Markdown(f"## {title}")
            gr.Plot(fig)

        gr.Markdown("## Interactive Maps")
        for label, file in maps.items():
            gr.Markdown(f"### {label}")
            with open(file, 'r', encoding='utf-8') as f:
                html = f.read().replace("'", "&#39;").replace("\n", " ")
                gr.HTML(f"<iframe srcdoc='{html}' width='100%' height='500' style='border:none;'></iframe>")

        gr.Markdown("## 🧠 Insights & Recommendations")
        gr.Markdown(insights)

    demo.launch()

launch_dashboard()