Update app.py

app.py

@@ -1,67 +1,220 @@
-
 import os
+import json
 import subprocess
+import urllib.request
+import urllib.error
+
 import gradio as gr
 import pandas as pd
+import numpy as np
 import joblib
 
-MODEL_PATH="models/pipeline.joblib"
+MODEL_PATH = "models/pipeline.joblib"
+PY_NOTEBOOK = "BankChurn_Version1.ipynb"
+R_NOTEBOOK = "BankChurn_Version1_R.ipynb"
 
+
+# =========================
+# Model
+# =========================
 def load_model():
     if os.path.exists(MODEL_PATH):
         return joblib.load(MODEL_PATH)
     return None
 
-model=load_model()
 
-def predict(age,balance):
-    global model
-    model=load_model()
+def predict(age, balance):
+    model = load_model()
     if model is None:
-        return "Run pipeline first"
-    df=pd.DataFrame([[age,balance]],columns=["Age","Balance"])
-    p=model.predict(df)[0]
-    return f"Prediction: {p}"
+        return "Please run the pipeline first."
+    df = pd.DataFrame([[age, balance]], columns=["Age", "Balance"])
+    pred = model.predict(df)[0]
+    return "Churn Risk: Yes" if pred == 1 else "Churn Risk: No"
+
 
 def run_pipeline():
-    proc=subprocess.Popen(
-        ["python","scripts/pipeline.py"],
+    proc = subprocess.Popen(
+        ["python", "scripts/pipeline.py"],
         stdout=subprocess.PIPE,
         stderr=subprocess.STDOUT,
         text=True
     )
-    log=""
+    log = ""
     for line in proc.stdout:
-        log+=line
+        log += line
         yield log
 
-def build_ui():
-    css=open("style.css").read()
 
-    with gr.Blocks() as demo:
+# =========================
+# Demo data for dashboard
+# =========================
+def make_demo_data():
+    np.random.seed(42)
+    n = 120
+    df = pd.DataFrame({
+        "CustomerID": range(1, n + 1),
+        "Age": np.random.randint(18, 70, n),
+        "Balance": np.random.randint(500, 10000, n),
+        "Tenure": np.random.randint(1, 10, n),
+        "Products": np.random.randint(1, 5, n),
+        "Geography": np.random.choice(["France", "Germany", "Spain"], n),
+        "Exited": np.random.choice([0, 1], n, p=[0.78, 0.22])
+    })
+    return df
+
+
+demo_df = make_demo_data()
+
+geo_df = demo_df.groupby("Geography", as_index=False)["Exited"].mean()
+geo_df["Exited"] = (geo_df["Exited"] * 100).round(2)
+
+age_df = demo_df.groupby(pd.cut(demo_df["Age"], bins=[18, 30, 40, 50, 60, 70])).agg(
+    churn_rate=("Exited", "mean")
+).reset_index()
+age_df["AgeBand"] = age_df["Age"].astype(str)
+age_df["churn_rate"] = (age_df["churn_rate"] * 100).round(2)
+
+summary_md = f"""
+### Dashboard Summary
+- Total Customers: **{len(demo_df)}**
+- Churned Customers: **{int(demo_df['Exited'].sum())}**
+- Churn Rate: **{round(demo_df['Exited'].mean() * 100, 2)}%**
+- Avg Balance: **${round(demo_df['Balance'].mean(), 2)}**
+"""
+
+
+# =========================
+# Notebook preview
+# =========================
+def load_notebook_preview(path, max_cells=6):
+    if not os.path.exists(path):
+        return "File not found."
+    with open(path, "r", encoding="utf-8") as f:
+        nb = json.load(f)
+
+    parts = []
+    for i, cell in enumerate(nb.get("cells", [])[:max_cells]):
+        source = "".join(cell.get("source", []))
+        parts.append(f"# Cell {i+1} ({cell.get('cell_type','code')})\n{source}\n")
+    return "\n\n".join(parts)
+
+
+py_preview = load_notebook_preview(PY_NOTEBOOK)
+r_preview = load_notebook_preview(R_NOTEBOOK)
 
+
+# =========================
+# HF AI integration
+# =========================
+def hf_ai_insight(question):
+    api_key = os.getenv("HF_API_KEY")
+    if not api_key:
+        return "HF_API_KEY not found. Add it in Space Secrets first."
+
+    prompt = f"""
+You are a banking analytics assistant.
+Context:
+- This app is about bank churn prediction.
+- Give short, practical business insights.
+User question: {question}
+"""
+
+    payload = json.dumps({"inputs": prompt}).encode("utf-8")
+    req = urllib.request.Request(
+        "https://api-inference.huggingface.co/models/google/flan-t5-base",
+        data=payload,
+        headers={
+            "Authorization": f"Bearer {api_key}",
+            "Content-Type": "application/json"
+        }
+    )
+
+    try:
+        with urllib.request.urlopen(req, timeout=60) as resp:
+            result = json.loads(resp.read().decode("utf-8"))
+        if isinstance(result, list) and len(result) > 0 and "generated_text" in result[0]:
+            return result[0]["generated_text"]
+        return str(result)
+    except urllib.error.HTTPError as e:
+        return f"HF API error: {e.read().decode('utf-8')}"
+    except Exception as e:
+        return f"Request failed: {str(e)}"
+
+
+# =========================
+# UI
+# =========================
+def build_ui():
+    css = open("style.css", "r", encoding="utf-8").read()
+
+    with gr.Blocks(title="Bank Churn Dashboard") as demo:
         gr.HTML(f"<style>{css}</style>")
 
-        gr.Markdown("# 🏦 Bank Churn Predictor")
+        gr.Markdown("""
+        # 🏦 Bank Churn Dashboard
+        Interactive churn analysis, model pipeline, prediction, and AI insight.
+        """)
+
+        with gr.Tab("Dashboard"):
+            gr.Markdown(summary_md)
+
+            with gr.Row():
+                churn_geo = gr.BarPlot(
+                    value=geo_df,
+                    x="Geography",
+                    y="Exited",
+                    title="Interactive Churn Rate by Geography (%)"
+                )
+                churn_age = gr.LinePlot(
+                    value=age_df,
+                    x="AgeBand",
+                    y="churn_rate",
+                    title="Interactive Churn Rate by Age Band (%)"
+                )
+
+            gr.Dataframe(
+                value=demo_df,
+                interactive=True,
+                label="Interactive Customer Table"
+            )
 
         with gr.Tab("Pipeline"):
-            gr.Markdown("Train model and view execution log")
-            btn=gr.Button("Run Pipeline")
-            log=gr.Textbox(lines=20,label="Execution Log")
-            btn.click(run_pipeline,outputs=log)
+            gr.Markdown("Train the model and inspect the execution log.")
+            btn_run = gr.Button("Run Pipeline")
+            log = gr.Textbox(lines=18, label="Execution Log")
+            btn_run.click(run_pipeline, outputs=log)
 
         with gr.Tab("Prediction"):
-            age=gr.Number(label="Age")
-            balance=gr.Number(label="Balance")
-            btn=gr.Button("Predict")
-            out=gr.Textbox()
+            age = gr.Number(label="Age", value=35)
+            balance = gr.Number(label="Balance", value=5000)
+            btn_pred = gr.Button("Predict Churn")
+            pred_out = gr.Textbox(label="Prediction Result")
+            btn_pred.click(predict, inputs=[age, balance], outputs=pred_out)
 
-            btn.click(predict,[age,balance],out)
+        with gr.Tab("Analysis Files"):
+            gr.Markdown("### Python analysis notebook")
+            gr.File(value=PY_NOTEBOOK, label="Download Python Notebook")
+            gr.Code(value=py_preview, language="python", label="Python Notebook Preview")
+
+            gr.Markdown("### R analysis notebook")
+            gr.File(value=R_NOTEBOOK, label="Download R Notebook")
+            gr.Code(value=r_preview, language="r", label="R Notebook Preview")
+
+        with gr.Tab("AI Insight"):
+            gr.Markdown("Ask AI for a churn insight. Requires `HF_API_KEY` in Space Secrets.")
+            ai_q = gr.Textbox(
+                label="Ask something",
+                placeholder="Example: What customer segment should the bank focus on retaining first?"
+            )
+            ai_btn = gr.Button("Generate AI Insight")
+            ai_out = gr.Textbox(lines=8, label="AI Response")
+            ai_btn.click(hf_ai_insight, inputs=ai_q, outputs=ai_out)
 
     return demo
 
-if __name__=="__main__":
-    demo=build_ui()
+
+if __name__ == "__main__":
+    demo = build_ui()
     demo.queue()
-    port=int(os.environ.get("PORT",7860))
-    demo.launch(server_name="0.0.0.0",server_port=port)
+    port = int(os.environ.get("PORT", 7860))
+    demo.launch(server_name="0.0.0.0", server_port=port)