Upload 5 files

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+airbnb_recommendation_output.csv filter=lfs diff=lfs merge=lfs -text
+synthetic_airbnb_project_data.csv filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,12 +1,49 @@
 ---
-title: StayWiseAI
-emoji: 🏃
-colorFrom: green
-colorTo: green
+title: AI Rental Performance Assistant
+emoji: 🏡
+colorFrom: blue
+colorTo: indigo
 sdk: gradio
-sdk_version: 6.14.0
+sdk_version: 4.44.0
 app_file: app.py
 pinned: false
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# AI Rental Performance Assistant
+
+This Hugging Face Space runs an end-to-end short-term rental pricing and performance pipeline.
+
+## What the app does
+
+1. Loads the Airbnb project datasets.
+2. Lets the user input a rental listing scenario.
+3. Finds comparable listings by neighbourhood group, neighbourhood, and room type.
+4. Calculates competitor price, price gap, estimated occupancy, revenue, demand score, and opportunity score.
+5. Produces a pricing recommendation.
+6. Sends the pipeline output to an n8n webhook.
+7. Displays the n8n response as an automation output.
+
+## n8n integration
+
+The app supports two ways to configure n8n:
+
+- Add your webhook URL directly in the app input field.
+- Or set a Hugging Face Space secret named `N8N_WEBHOOK_URL`.
+
+Expected n8n response format:
+
+```json
+{
+  "status": "success",
+  "insight": "This listing has high demand potential but is underpriced.",
+  "next_step": "Consider increasing price by 8% and improving review quality.",
+  "log": "Saved to database and flagged for monitoring"
+}
+```
+
+## Files
+
+- `app.py`: main Gradio app
+- `requirements.txt`: Python dependencies
+- `synthetic_airbnb_project_data.csv`: full project dataset
+- `airbnb_recommendation_output.csv`: recommendation output dataset

airbnb_recommendation_output.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18cce6473679041361120d96e14af3742f994175b8387c9a90b5a65328f3fa14
+size 10814149

app.py

@@ -0,0 +1,373 @@
+import os
+import json
+from datetime import datetime
+from typing import Dict, Tuple
+
+import numpy as np
+import pandas as pd
+import plotly.express as px
+import requests
+import gradio as gr
+
+DATA_FULL = "synthetic_airbnb_project_data.csv"
+DATA_RECO = "airbnb_recommendation_output.csv"
+
+# -----------------------------
+# Data loading
+# -----------------------------
+
+@gr.cache()
+def load_data() -> Tuple[pd.DataFrame, pd.DataFrame]:
+    full = pd.read_csv(DATA_FULL)
+    reco = pd.read_csv(DATA_RECO)
+
+    # Defensive cleaning
+    for df in (full, reco):
+        if "price" in df.columns:
+            df["price"] = pd.to_numeric(df["price"], errors="coerce")
+        if "occupancy_rate" in df.columns:
+            df["occupancy_rate"] = pd.to_numeric(df["occupancy_rate"], errors="coerce")
+        if "monthly_revenue" in df.columns:
+            df["monthly_revenue"] = pd.to_numeric(df["monthly_revenue"], errors="coerce")
+        if "demand_score" in df.columns:
+            df["demand_score"] = pd.to_numeric(df["demand_score"], errors="coerce")
+        if "customer_sentiment_score" in df.columns:
+            df["customer_sentiment_score"] = pd.to_numeric(df["customer_sentiment_score"], errors="coerce")
+        if "synthetic_rating" in df.columns:
+            df["synthetic_rating"] = pd.to_numeric(df["synthetic_rating"], errors="coerce")
+
+    full = full.dropna(subset=["price", "neighbourhood_group", "room_type"])
+    reco = reco.dropna(subset=["price", "neighbourhood_group", "room_type"])
+    return full, reco
+
+full_df, reco_df = load_data()
+
+NEIGH_GROUPS = sorted(full_df["neighbourhood_group"].dropna().unique().tolist())
+ROOM_TYPES = sorted(full_df["room_type"].dropna().unique().tolist())
+SEASONS = sorted(full_df["season"].dropna().unique().tolist()) if "season" in full_df.columns else ["Spring", "Summer", "Autumn", "Winter"]
+
+
+def neighbourhood_choices(group: str):
+    if not group:
+        return gr.Dropdown(choices=[], value=None)
+    vals = sorted(full_df.loc[full_df["neighbourhood_group"] == group, "neighbourhood"].dropna().unique().tolist())
+    return gr.Dropdown(choices=vals, value=vals[0] if vals else None)
+
+
+# -----------------------------
+# Pipeline logic
+# -----------------------------
+
+def safe_mean(series, default=0):
+    val = pd.to_numeric(series, errors="coerce").mean()
+    if pd.isna(val):
+        return default
+    return float(val)
+
+
+def get_comparables(group: str, neighbourhood: str, room_type: str, price: float) -> pd.DataFrame:
+    # 1) Most precise: same neighbourhood + room type
+    comps = full_df[(full_df["neighbourhood_group"] == group) &
+                    (full_df["neighbourhood"] == neighbourhood) &
+                    (full_df["room_type"] == room_type)].copy()
+
+    # 2) Fallback: same group + room type
+    if len(comps) < 10:
+        comps = full_df[(full_df["neighbourhood_group"] == group) &
+                        (full_df["room_type"] == room_type)].copy()
+
+    # 3) Fallback: same room type globally
+    if len(comps) < 10:
+        comps = full_df[full_df["room_type"] == room_type].copy()
+
+    # 4) Prefer listings in a reasonable price band, but only if it does not remove too much data
+    if price and price > 0 and len(comps) > 30:
+        band = comps[(comps["price"] >= price * 0.5) & (comps["price"] <= price * 1.5)]
+        if len(band) >= 10:
+            comps = band
+
+    return comps
+
+
+def demand_level_from_score(score: float) -> str:
+    if score >= 70:
+        return "High"
+    if score >= 45:
+        return "Medium"
+    return "Low"
+
+
+def build_recommendation(price, competitor_avg, occupancy, demand_score, sentiment, rating):
+    price_gap_pct = ((price - competitor_avg) / competitor_avg * 100) if competitor_avg else 0
+
+    if price_gap_pct > 18 and demand_score < 65:
+        pricing_action = "Consider lowering price"
+        suggested_price = round(competitor_avg * 1.05, 0)
+    elif price_gap_pct < -12 and demand_score >= 60 and occupancy >= 0.60:
+        pricing_action = "Consider raising price"
+        suggested_price = round(min(competitor_avg * 0.98, price * 1.12), 0)
+    else:
+        pricing_action = "Keep price stable"
+        suggested_price = round(price, 0)
+
+    actions = []
+    if pricing_action == "Consider lowering price":
+        actions.append("Lower price slightly to recover competitiveness")
+    elif pricing_action == "Consider raising price":
+        actions.append("Raise price carefully because demand is strong")
+    else:
+        actions.append("Keep price stable and monitor demand")
+
+    if occupancy < 0.45:
+        actions.append("Boost visibility or add promotions")
+    elif occupancy > 0.70:
+        actions.append("Strong occupancy: test premium pricing")
+    else:
+        actions.append("Demand is stable")
+
+    if sentiment < 0:
+        actions.append("Improve guest experience based on negative sentiment")
+    elif rating < 4.2:
+        actions.append("Improve rating drivers such as cleanliness, check-in, and value")
+    else:
+        actions.append("Satisfaction looks healthy")
+
+    opportunity_score = round((demand_score * 0.45) + (occupancy * 100 * 0.25) + (max(sentiment, -1) + 1) * 15 + (rating / 5) * 15, 2)
+    return pricing_action, suggested_price, price_gap_pct, " | ".join(actions), opportunity_score
+
+
+def send_to_n8n(payload: Dict, webhook_url: str) -> Dict:
+    if not webhook_url:
+        webhook_url = os.getenv("N8N_WEBHOOK_URL", "").strip()
+
+    if not webhook_url:
+        # Demo response so the Space works even before n8n is connected.
+        action = payload.get("pricing_recommendation", "Keep price stable")
+        if action == "Consider raising price":
+            insight = "This listing has strong demand potential and may be underpriced compared with market conditions."
+            next_step = f"Test the suggested price of ${payload.get('suggested_price', payload.get('price'))} and monitor occupancy."
+        elif action == "Consider lowering price":
+            insight = "This listing appears less competitive on price compared with similar listings."
+            next_step = "Lower the price slightly, improve listing visibility, and track whether bookings improve."
+        else:
+            insight = "This listing is reasonably aligned with comparable market conditions."
+            next_step = "Keep price stable, monitor demand, and focus on guest experience improvements."
+
+        return {
+            "status": "demo_mode",
+            "insight": insight,
+            "next_step": next_step,
+            "log": "No n8n webhook configured. Demo automation response generated inside the app."
+        }
+
+    try:
+        response = requests.post(webhook_url, json=payload, timeout=15)
+        response.raise_for_status()
+        try:
+            data = response.json()
+        except Exception:
+            data = {"status": "success", "insight": response.text[:300], "next_step": "Review workflow output.", "log": "n8n returned plain text."}
+        return data
+    except Exception as e:
+        return {
+            "status": "error",
+            "insight": "The pricing pipeline ran, but the n8n automation did not complete.",
+            "next_step": "Check the webhook URL, n8n workflow activation, and Respond to Webhook node.",
+            "log": str(e)
+        }
+
+
+def run_full_pipeline(group, neighbourhood, room_type, price, availability_365, season, local_event_score, rating, sentiment, webhook_url):
+    price = float(price)
+    availability_365 = int(availability_365)
+    local_event_score = float(local_event_score)
+    rating = float(rating)
+    sentiment = float(sentiment)
+
+    comps = get_comparables(group, neighbourhood, room_type, price)
+
+    competitor_avg = safe_mean(comps["price"], default=price)
+    base_occupancy = safe_mean(comps["occupancy_rate"], default=0.50)
+    base_demand = safe_mean(comps["demand_score"], default=50)
+    base_revenue = safe_mean(comps["monthly_revenue"], default=price * 15)
+
+    # Adjusted demand logic: interpretable, stable, and presentation-friendly.
+    price_gap = ((price - competitor_avg) / competitor_avg * 100) if competitor_avg else 0
+    availability_factor = max(0, min(1, (365 - availability_365) / 365))
+    event_boost = local_event_score * 2.2
+    sentiment_boost = sentiment * 8
+    rating_boost = (rating - 4.0) * 8
+    price_penalty = max(0, price_gap) * 0.25
+
+    demand_score = float(np.clip(base_demand * 0.45 + availability_factor * 25 + event_boost + sentiment_boost + rating_boost - price_penalty, 0, 100))
+    demand_level = demand_level_from_score(demand_score)
+
+    occupancy = float(np.clip(base_occupancy * 0.55 + availability_factor * 0.25 + (demand_score / 100) * 0.20, 0.05, 0.95))
+    booked_nights = round(occupancy * 30)
+    monthly_revenue = round(price * booked_nights, 0)
+
+    pricing_action, suggested_price, price_gap_pct, final_reco, opportunity_score = build_recommendation(
+        price, competitor_avg, occupancy, demand_score, sentiment, rating
+    )
+
+    payload = {
+        "timestamp": datetime.utcnow().isoformat() + "Z",
+        "neighbourhood_group": group,
+        "neighbourhood": neighbourhood,
+        "room_type": room_type,
+        "season": season,
+        "price": price,
+        "suggested_price": suggested_price,
+        "competitor_avg_price": round(competitor_avg, 2),
+        "price_vs_competitor_pct": round(price_gap_pct, 2),
+        "availability_365": availability_365,
+        "local_event_score": local_event_score,
+        "synthetic_rating": rating,
+        "customer_sentiment_score": sentiment,
+        "estimated_occupancy_rate": round(occupancy, 3),
+        "estimated_booked_nights_month": booked_nights,
+        "estimated_monthly_revenue": monthly_revenue,
+        "demand_score": round(demand_score, 1),
+        "demand_level": demand_level,
+        "pricing_recommendation": pricing_action,
+        "final_recommendation": final_reco,
+        "opportunity_score": opportunity_score,
+        "comparable_listing_count": int(len(comps))
+    }
+
+    n8n_response = send_to_n8n(payload, webhook_url.strip() if webhook_url else "")
+
+    cards = f"""
+    <div class='cards'>
+      <div class='card'><div class='label'>Pricing Action</div><div class='value'>{pricing_action}</div><div class='sub'>Suggested price: ${suggested_price:,.0f}</div></div>
+      <div class='card'><div class='label'>Demand Level</div><div class='value'>{demand_level}</div><div class='sub'>Demand score: {demand_score:.1f}/100</div></div>
+      <div class='card'><div class='label'>Monthly Revenue</div><div class='value'>${monthly_revenue:,.0f}</div><div class='sub'>{booked_nights} booked nights estimated</div></div>
+      <div class='card'><div class='label'>Opportunity Score</div><div class='value'>{opportunity_score:.1f}</div><div class='sub'>{len(comps)} comparable listings</div></div>
+    </div>
+    """
+
+    explanation = f"""
+    ### Strategic Recommendation
+
+    **{final_reco}**
+
+    - Your entered price is **${price:,.0f}**.
+    - Comparable listings average **${competitor_avg:,.0f}**.
+    - Your price is **{price_gap_pct:+.1f}%** versus comparable listings.
+    - Estimated occupancy is **{occupancy:.1%}**.
+    - Estimated monthly revenue is **${monthly_revenue:,.0f}**.
+    - Demand is classified as **{demand_level}**.
+    """
+
+    automation = f"""
+    ### n8n Automation Output
+
+    **Status:** `{n8n_response.get('status', 'unknown')}`
+
+    **Insight:** {n8n_response.get('insight', 'No insight returned.')}
+
+    **Next step:** {n8n_response.get('next_step', 'No next step returned.')}
+
+    **Log:** {n8n_response.get('log', 'No log returned.')}
+    """
+
+    # Charts
+    comp_sample = comps[["name", "neighbourhood", "room_type", "price", "occupancy_rate", "monthly_revenue", "demand_score"]].copy()
+    comp_sample = comp_sample.sort_values("price").head(100)
+
+    fig_price = px.histogram(comps, x="price", nbins=40, title="Comparable Listing Price Distribution")
+    fig_price.add_vline(x=price, line_dash="dash", annotation_text="Your price", annotation_position="top")
+
+    scatter_cols = ["price", "demand_score", "room_type", "neighbourhood"]
+    fig_demand = px.scatter(comps.dropna(subset=["price", "demand_score"]), x="price", y="demand_score", color="room_type",
+                            hover_data=["neighbourhood"], title="Price vs Demand Score for Comparable Listings")
+
+    table = comp_sample.head(12).round(2)
+    payload_pretty = json.dumps(payload, indent=2)
+
+    return cards, explanation, automation, fig_price, fig_demand, table, payload_pretty
+
+
+# -----------------------------
+# UI
+# -----------------------------
+
+CSS = """
+.gradio-container {max-width: 1280px !important; margin: auto;}
+.hero {
+  background: linear-gradient(135deg, #0f172a 0%, #1e3a8a 50%, #2563eb 100%);
+  border-radius: 24px;
+  padding: 34px;
+  color: white;
+  box-shadow: 0 18px 45px rgba(15, 23, 42, 0.22);
+  margin-bottom: 18px;
+}
+.hero h1 {font-size: 38px; margin: 0 0 8px 0;}
+.hero p {font-size: 16px; opacity: 0.92; margin: 0; max-width: 900px;}
+.cards {display: grid; grid-template-columns: repeat(4, minmax(0, 1fr)); gap: 14px; margin: 10px 0 18px 0;}
+.card {background: white; border: 1px solid #e5e7eb; border-radius: 18px; padding: 18px; box-shadow: 0 10px 25px rgba(15, 23, 42, 0.08);}
+.label {font-size: 13px; text-transform: uppercase; color: #64748b; letter-spacing: .04em;}
+.value {font-size: 24px; font-weight: 800; color: #0f172a; margin-top: 8px;}
+.sub {font-size: 13px; color: #475569; margin-top: 8px;}
+.panel-note {background: #eff6ff; border-left: 5px solid #2563eb; padding: 12px 14px; border-radius: 12px; color: #1e3a8a;}
+@media (max-width: 900px) {.cards {grid-template-columns: repeat(2, minmax(0, 1fr));}}
+@media (max-width: 600px) {.cards {grid-template-columns: 1fr;}.hero h1{font-size:28px;}}
+"""
+
+with gr.Blocks(css=CSS, theme=gr.themes.Soft(primary_hue="blue", neutral_hue="slate")) as demo:
+    gr.HTML("""
+    <div class='hero'>
+      <h1>🏡 AI Rental Performance Assistant</h1>
+      <p>Run a full short-term rental pricing pipeline: comparable listings, demand scoring, revenue estimation, pricing recommendation, and n8n automation response.</p>
+    </div>
+    """)
+
+    with gr.Row():
+        with gr.Column(scale=1):
+            gr.Markdown("### 1) Listing Inputs")
+            group = gr.Dropdown(choices=NEIGH_GROUPS, value=NEIGH_GROUPS[0] if NEIGH_GROUPS else None, label="Neighbourhood Group")
+            neighbourhood = gr.Dropdown(choices=sorted(full_df.loc[full_df["neighbourhood_group"] == (NEIGH_GROUPS[0] if NEIGH_GROUPS else ""), "neighbourhood"].dropna().unique().tolist()), label="Neighbourhood")
+            room_type = gr.Dropdown(choices=ROOM_TYPES, value=ROOM_TYPES[0] if ROOM_TYPES else None, label="Room Type")
+            price = gr.Slider(20, 800, value=150, step=5, label="Current Nightly Price ($)")
+            availability_365 = gr.Slider(0, 365, value=180, step=1, label="Availability Over Next 365 Days")
+            season = gr.Dropdown(choices=SEASONS, value=SEASONS[0] if SEASONS else "Summer", label="Season")
+            local_event_score = gr.Slider(0, 10, value=5, step=0.1, label="Local Event Score (0 = low, 10 = major event period)")
+            rating = gr.Slider(1, 5, value=4.4, step=0.05, label="Guest Rating")
+            sentiment = gr.Slider(-1, 1, value=0.25, step=0.01, label="Customer Sentiment Score")
+
+            gr.Markdown("### 2) n8n Webhook")
+            gr.HTML("<div class='panel-note'>Paste your n8n production webhook URL here, or leave blank to run demo mode.</div>")
+            webhook_url = gr.Textbox(label="n8n Webhook URL", placeholder="https://your-n8n-domain/webhook/airbnb-pipeline", type="password")
+            run_btn = gr.Button("🚀 Run Full Pipeline", variant="primary", size="lg")
+
+        with gr.Column(scale=2):
+            gr.Markdown("### 3) Pipeline Output")
+            cards = gr.HTML()
+            explanation = gr.Markdown()
+            automation = gr.Markdown()
+
+    with gr.Tabs():
+        with gr.Tab("📊 Charts"):
+            with gr.Row():
+                fig_price = gr.Plot()
+                fig_demand = gr.Plot()
+        with gr.Tab("🏘️ Comparable Listings"):
+            comp_table = gr.Dataframe(label="Top comparable listings", interactive=False)
+        with gr.Tab("🧾 JSON Sent to n8n"):
+            payload_view = gr.Code(language="json", label="Pipeline payload")
+
+    group.change(fn=neighbourhood_choices, inputs=group, outputs=neighbourhood)
+
+    run_btn.click(
+        fn=run_full_pipeline,
+        inputs=[group, neighbourhood, room_type, price, availability_365, season, local_event_score, rating, sentiment, webhook_url],
+        outputs=[cards, explanation, automation, fig_price, fig_demand, comp_table, payload_view]
+    )
+
+    gr.Markdown("""
+    ---
+    **Project logic:** The app finds comparable listings, estimates demand and revenue, produces a pricing recommendation, then sends the full result to n8n. n8n should return a JSON response with `status`, `insight`, `next_step`, and `log`.
+    """)
+
+if __name__ == "__main__":
+    demo.launch()

requirements.txt

@@ -0,0 +1,5 @@
+gradio>=4.44.0
+pandas>=2.0.0
+numpy>=1.24.0
+plotly>=5.18.0
+requests>=2.31.0

synthetic_airbnb_project_data.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e63f75596406a07ad35fb85e63a803708244e352792c6dcdf323bf788c00b19
+size 17968855