Create app.py

app.py

@@ -0,0 +1,166 @@
+import gradio as gr
+import pandas as pd
+import matplotlib.pyplot as plt
+import matplotlib
+matplotlib.use("Agg")
+import seaborn as sns
+import requests
+import io
+from pathlib import Path
+from vaderSentiment.vaderSentiment import SentimentIntensityAnalyzer
+
+# ─────────────────────────────────────────
+# SECTION 1 — Charger les données pré-calculées
+# ─────────────────────────────────────────
+try:
+    df_pricing   = pd.read_csv("artifacts/pricing_decisions.csv")
+    df_sales     = pd.read_csv("artifacts/dashboard_data.csv")
+    ARTIFACTS_OK = True
+except Exception:
+    ARTIFACTS_OK = False
+    df_pricing   = pd.DataFrame()
+    df_sales     = pd.DataFrame()
+
+# ─────────────────────────────────────────
+# SECTION 2 — Analyse en temps réel (VADER)
+# ─────────────────────────────────────────
+analyzer = SentimentIntensityAnalyzer()
+
+def get_sentiment_label(text):
+    score = analyzer.polarity_scores(text)["compound"]
+    if score >= 0.05:
+        return "positive"
+    elif score <= -0.05:
+        return "negative"
+    else:
+        return "neutral"
+
+def pricing_decision(avg_units, positive_ratio, negative_ratio):
+    if avg_units >= 120 and positive_ratio >= 0.6:
+        return "📈 Increase Price"
+    elif avg_units <= 60 and negative_ratio >= 0.4:
+        return "📉 Decrease Price"
+    else:
+        return "➡️ Keep Price"
+
+def analyze_book(title, reviews_text, avg_units_sold):
+    if not title or not reviews_text:
+        return "⚠️ Please enter a title and at least one review.", "", None
+
+    # Analyse sentiment de chaque review
+    lines = [r.strip() for r in reviews_text.strip().split("\n") if r.strip()]
+    labels = [get_sentiment_label(line) for line in lines]
+
+    total         = len(labels)
+    positive_ratio = labels.count("positive") / total
+    negative_ratio = labels.count("negative") / total
+    neutral_ratio  = labels.count("neutral")  / total
+
+    decision = pricing_decision(avg_units_sold, positive_ratio, negative_ratio)
+
+    # Résumé texte
+    summary = f"""
+📚 **{title}**
+
+🔢 Reviews analysées : {total}
+😊 Positive : {positive_ratio:.0%}
+😐 Neutral  : {neutral_ratio:.0%}
+😞 Negative : {negative_ratio:.0%}
+📦 Avg units sold : {avg_units_sold}
+
+💡 **Pricing Decision : {decision}**
+"""
+
+    # Graphique en camembert
+    fig, ax = plt.subplots(figsize=(4, 4))
+    colors = ["#4CAF50", "#FFC107", "#F44336"]
+    ax.pie(
+        [positive_ratio, neutral_ratio, negative_ratio],
+        labels=["Positive", "Neutral", "Negative"],
+        autopct="%1.0f%%",
+        colors=colors,
+        startangle=90
+    )
+    ax.set_title(f"Sentiment — {title}")
+    plt.tight_layout()
+
+    return summary, "\n".join(f"{l} → {s}" for l, s in zip(lines, labels)), fig
+
+# ─────────────────────────────────────────
+# SECTION 3 — Interface Gradio
+# ─────────────────────────────────────────
+with gr.Blocks(title="📚 Book Price Decider", theme=gr.themes.Soft()) as app:
+
+    gr.Markdown("# 📚 Book Price Decider — Group A4")
+    gr.Markdown("Sentiment analysis + ARIMA-based pricing decisions for books.")
+
+    with gr.Tabs():
+
+        # ── Tab 1 : Dashboard pré-calculé ──────────────────
+        with gr.Tab("📊 Dashboard"):
+            gr.Markdown("### Pre-computed results from the analysis notebooks")
+
+            if ARTIFACTS_OK:
+                with gr.Row():
+                    gr.Image(value="artifacts/sales_trends.png",
+                             label="Sales Trends")
+                    gr.Image(value="artifacts/sentiment_distribution.png",
+                             label="Sentiment Distribution")
+                gr.Dataframe(value=df_pricing, label="Pricing Decisions Table")
+            else:
+                gr.Markdown(
+                    "⚠️ No artifacts found yet. "
+                    "Run the notebooks and upload the `artifacts/` folder."
+                )
+
+        # ── Tab 2 : Analyse en temps réel ──────────────────
+        with gr.Tab("🔮 Analyze a New Book"):
+            gr.Markdown("### Enter book info to get a live pricing recommendation")
+
+            with gr.Row():
+                title_input = gr.Textbox(label="Book Title", placeholder="e.g. The Great Gatsby")
+                units_input = gr.Number(label="Avg Monthly Units Sold", value=100)
+
+            reviews_input = gr.Textbox(
+                label="Paste reviews here (one per line)",
+                lines=6,
+                placeholder="This book was amazing!\nNot what I expected.\nDecent read overall."
+            )
+
+            analyze_btn = gr.Button("🚀 Analyze & Decide", variant="primary")
+
+            with gr.Row():
+                summary_output  = gr.Markdown(label="Summary")
+                details_output  = gr.Textbox(label="Review-by-review labels", lines=6)
+
+            chart_output = gr.Plot(label="Sentiment Chart")
+
+            analyze_btn.click(
+                fn=analyze_book,
+                inputs=[title_input, reviews_input, units_input],
+                outputs=[summary_output, details_output, chart_output]
+            )
+
+        # ── Tab 3 : À propos ───────────────────────────────
+        with gr.Tab("ℹ️ About"):
+            gr.Markdown("""
+## About this app
+
+This app is part of the **AI for Big Data Management** group project at ESCP Business School.
+
+### Pipeline
+1. **Real-world data** scraped from Books to Scrape
+2. **Synthetic data** generated to enrich with reviews & sales history
+3. **VADER sentiment analysis** on customer reviews
+4. **ARIMA forecasting** on sales time series
+5. **Rule-based pricing decisions** combining sentiment + sales volume
+6. **This Hugging Face app** as the final automation layer
+
+### Team — Group A4
+- Project Manager
+- Data Analyst
+- UX Designer(s)
+- Content Specialist
+""")
+
+app.launch()