feat: Enhance risk analysis with new risk drivers, update UI styling, and add comprehensive reporting with visual analytics and integrated AI advice.

app.py

@@ -6,147 +6,190 @@ import plotly.graph_objects as go
 from google import genai
 from datetime import timedelta
 
-# --- CONFIG & AI ---
+# --- INITIALIZATION ---
 GOOGLE_API_KEY = os.getenv("GOOGLE_API_KEY")
-client_ai = genai.Client(api_key=GOOGLE_API_KEY) if GOOGLE_API_KEY else None
+ai_client = genai.Client(api_key=GOOGLE_API_KEY) if GOOGLE_API_KEY else None
 
-# --- UI STYLING ---
+# --- UI STYLE: BANK NAGARI PREMIUM ---
+# Palette: #0514DE (Deep Blue), #82C3EB (Light Blue), #F7BD87 (Gold), #FFFFFF (White)
 custom_css = """
 @import url('https://fonts.googleapis.com/css2?family=Plus+Jakarta+Sans:wght@400;600;700;800&display=swap');
-body, .gradio-container { font-family: 'Plus Jakarta Sans', sans-serif !important; background-color: #FFFFFF !important; }
-.nagari-header { background: linear-gradient(135deg, #0514DE 0%, #82C3EB 100%); padding: 35px; border-radius: 15px; border-bottom: 6px solid #F7BD87; margin-bottom: 25px; text-align: center; }
-.nagari-header h1 { color: #FFFFFF !important; font-weight: 800 !important; margin: 0; }
-.card { background: #FFFFFF; border-radius: 12px; padding: 20px; border: 1px solid #E0EDF4; box-shadow: 0 4px 10px rgba(0,0,0,0.05); }
-.status-pill { padding: 5px 15px; border-radius: 20px; color: white; font-weight: bold; font-size: 0.9em; }
+body, .gradio-container { font-family: 'Plus Jakarta Sans', sans-serif !important; background-color: #f8fafc !important; }
+
+.nagari-header { 
+    background: linear-gradient(135deg, #0514DE 0%, #82C3EB 100%); 
+    padding: 35px; border-radius: 15px; border-bottom: 6px solid #F7BD87; 
+    margin-bottom: 25px; text-align: center; box-shadow: 0 10px 15px rgba(5, 20, 222, 0.15);
+}
+
+.nagari-header h1 { 
+    color: #FFFFFF !important; font-weight: 800 !important; margin: 0; font-size: 2.2em; letter-spacing: 1px;
+}
+
+.card-overview { background: #FFFFFF; border-radius: 12px; padding: 25px; border: 1px solid #e2e8f0; box-shadow: 0 1px 3px rgba(0,0,0,0.1); }
+
+.report-section { background: #FFFFFF; border-radius: 12px; padding: 30px; border-left: 8px solid #0514DE; min-height: 400px; }
+
+.status-badge { padding: 8px 20px; border-radius: 30px; color: white; font-weight: 700; display: inline-block; margin-bottom: 15px; }
+
+.advice-container { background: #fffdf0; border: 1px solid #F7BD87; padding: 20px; border-radius: 10px; margin-top: 20px; }
 """
 
-class ArchonBankNagari:
+class ArchonNagariEngine:
     def __init__(self):
         self.load_data()
 
     def load_data(self):
-        # FASE 1: DATA FOUNDATION
-        self.df_txn = pd.read_csv('transactions.csv', parse_dates=['date']).sort_values('date')
-        self.df_cust = pd.read_csv('customers.csv')
-        self.df_bal = pd.read_csv('balances_revised.csv', parse_dates=['month']).sort_values('month')
-        self.df_rep = pd.read_csv('repayments_revised.csv', parse_dates=['due_date'])
-        self.df_off = pd.read_csv('offers.csv')
+        # FASE 1: DATA FOUNDATION (Single Source of Truth)
+        try:
+            self.df_txn = pd.read_csv('transactions.csv', parse_dates=['date']).sort_values('date')
+            self.df_cust = pd.read_csv('customers.csv')
+            self.df_bal = pd.read_csv('balances_revised.csv', parse_dates=['month']).sort_values('month')
+            self.df_rep = pd.read_csv('repayments_revised.csv', parse_dates=['due_date'])
+        except Exception as e:
+            print(f"Data Loading Error: {e}")
 
     def analyze(self, customer_id):
         cid = str(customer_id).strip().upper()
         u_txn = self.df_txn[self.df_txn['customer_id'] == cid].copy()
-        u_bal = self.df_bal[self.df_bal['customer_id'] == cid].copy()
-        u_rep = self.df_rep[self.df_rep['customer_id'] == cid].copy()
-        u_info = self.df_cust[self.df_cust['customer_id'] == cid].iloc[0] if cid in self.df_cust['customer_id'].values else None
+        u_bal = self.df_bal[self.df_bal['customer_id'] == cid].sort_values('month')
+        u_rep = self.df_rep[self.df_rep['customer_id'] == cid]
+        u_info_list = self.df_cust[self.df_cust['customer_id'] == cid]
 
-        if u_txn.empty or u_info is None: return None
+        if u_txn.empty or u_info_df_empty := u_info_list.empty: return None
+        u_info = u_info_list.iloc[0]
 
-        # --- FASE 2: TRANSACTION INTELLIGENCE ---
-        # Mapping Expense Type
+        # --- FASE 2: TRANSACTION INTELLIGENCE (Non-ML Rules) ---
         essential_cats = {'groceries', 'utilities', 'transport', 'healthcare', 'education'}
         u_txn['expense_type'] = u_txn['raw_description'].apply(lambda x: 'essential' if any(k in x.lower() for k in essential_cats) else 'discretionary')
         
-        # Risk Spending Flag (Rolling 30 days median)
+        # Risk Spending Flag (Rolling 30D Median)
         u_txn = u_txn.set_index('date').sort_index()
         u_txn['rolling_median'] = u_txn['amount'].rolling('30D').median()
         u_txn['risk_spending_flag'] = ((u_txn['expense_type'] == 'discretionary') & (u_txn['amount'] > u_txn['rolling_median'])).astype(int)
         u_txn = u_txn.reset_index()
 
-        # Behavior Signal
-        q75 = u_txn['amount'].quantile(0.75)
-        def get_signal(row):
-            if row['expense_type'] == 'discretionary' and row['amount'] > q75: return 'impulsive'
-            return 'normal'
-        u_txn['behavior_signal'] = u_txn.apply(get_signal, axis=1)
-
-        # --- FASE 3 & 4: AGGREGATION & RISK ---
-        income = u_txn[u_txn['transaction_type'] == 'credit']['amount'].sum()
+        # --- FASE 3 & 4: RISK LABELING (30/20/20/20/10) ---
+        income_txn = u_txn[u_txn['transaction_type'] == 'credit']['amount'].sum()
+        ref_income = max(income_txn, u_info['monthly_income'])
         expense = u_txn[u_txn['transaction_type'] == 'debit']['amount'].sum()
-        ref_inc = max(income, u_info['monthly_income'])
-        er = min(expense / ref_inc, 1.0)
+        er = min(expense / ref_income, 1.0)
         
-        # Risk Scoring (30/20/20/20/10)
         er_s = 1.0 if er > 0.8 else (0.5 if er > 0.5 else 0.0)
         bt_s = 1.0 if len(u_bal) >= 2 and u_bal.iloc[-1]['avg_balance'] < u_bal.iloc[-2]['avg_balance'] else 0.0
         od_s = 1.0 if (u_bal['min_balance'] <= 0).any() else 0.0
         mp_s = 1.0 if (u_rep['status'] == 'late').any() else 0.0
         
-        final_score = (0.3 * er_s) + (0.2 * bt_s) + (0.2 * od_s) + (0.2 * mp_s) + 0.1
-        risk_lv = "HIGH" if final_score >= 0.7 else ("MEDIUM" if final_score >= 0.4 else "LOW")
-
+        risk_score = (0.3 * er_s) + (0.2 * bt_s) + (0.2 * od_s) + (0.2 * mp_s) + 0.1
+        risk_lv = "HIGH" if risk_score >= 0.7 else ("MEDIUM" if risk_score >= 0.4 else "LOW")
+        
+        # Risk Drivers
+        drivers = []
+        if er_s >= 0.5: drivers.append("HIGH_EXPENSE_RATIO")
+        if bt_s == 1.0: drivers.append("DECLINING_BALANCE")
+        if od_s == 1.0: drivers.append("OVERDRAFT_HISTORY")
+        
         # --- FASE 5: NBO ENGINE ---
-        disc_ratio = u_txn[u_txn['expense_type'] == 'discretionary']['amount'].sum() / expense if expense > 0 else 0
+        disc_spending = u_txn[u_txn['expense_type'] == 'discretionary']['amount'].sum()
+        disc_ratio = disc_spending / expense if expense > 0 else 0
+        
         if risk_lv == "HIGH" or mp_s == 1:
             action, reason = "Restructuring Suggestion", "repayment_risk_detected"
-        elif er > 0.6:
+        elif er > 0.6 or disc_ratio > 0.5:
             action, reason = "Spending Control", "high_discretionary_spending"
-        elif risk_lv == "LOW" and disc_ratio < 0.3:
+        elif risk_lv == "LOW":
             action, reason = "Promote Investment", "surplus_balance"
         else:
             action, reason = "Financial Education", "stable_cashflow"
 
-        return risk_lv, final_score, er, u_bal, u_txn, action, reason
+        return risk_lv, risk_score, er, disc_ratio, u_bal, u_txn, expense, ref_income, action, reason, drivers
 
-    def get_ai_narrative(self, risk_lv, er, cid, u_txn):
-        if not client_ai: return "Koneksi Advisor AI tidak tersedia."
-        tx = u_txn.tail(2)['raw_description'].tolist()
-        prompt = f"Advisor Bank Nagari: Nasabah {cid} ({risk_lv} risk, expense {er:.1%}). Terakhir belanja di {tx}. Beri 1 saran hangat personal (Bapak/Ibu) maks 3 kalimat."
-        try:
-            return client_ai.models.generate_content(model="gemini-1.5-flash", contents=prompt).text
-        except: return "Kami menyarankan Bapak/Ibu untuk menjaga rasio pengeluaran agar tetap stabil bulan ini."
+    def get_summary_report(self, risk_lv, score, er, u_bal, expense, income, action, reason, drivers, cid, u_txn):
+        # FASE 6: EXPLAINABLE SUMMARY (CLEAN & PROFESSIONAL)
+        msg = f"### 📊 LAPORAN ANALISIS RESILIENSI: {risk_lv} RISK\n\n"
+        
+        msg += f"**1. Evaluasi Parameter Risiko (Fase 4)**\n"
+        msg += f"* **Indeks Risiko**: {score:.2f} (Skala 0-1). Perhitungan menggunakan bobot perbankan Nagari.\n"
+        msg += f"* **Faktor Pemicu (Drivers)**: {', '.join(drivers) if drivers else 'Normal'}\n"
+        msg += f"* **Efisiensi Anggaran**: {er:.1%}. Bapak/Ibu mengalokasikan Rp{expense:,.0f} dari total daya beli Rp{income:,.0f}.\n\n"
+        
+        msg += f"**2. Profil Kesehatan Saldo (Fase 3)**\n"
+        if not u_bal.empty:
+            last_bal = u_bal.iloc[-1]['avg_balance']
+            msg += f"* **Saldo Rata-rata**: Rp{last_bal:,.0f}.\n"
+            msg += f"* **Status Tren**: {'Terdeteksi tren penurunan saldo harian.' if len(u_bal) > 1 and last_bal < u_bal.iloc[-2]['avg_balance'] else 'Pertumbuhan saldo terpantau stabil.'}\n\n"
+        
+        msg += f"**3. Rekomendasi Tindakan (NBO Engine - Fase 5)**\n"
+        msg += f"* **Aksi Rekomendasi**: **{action}**\n"
+        msg += f"* **Dasar Keputusan**: {reason}\n"
+
+        # Adaptive AI Advice
+        if ai_client:
+            try:
+                tx_last = u_txn.tail(2)['raw_description'].tolist()
+                prompt = f"Advisor Bank Nagari: Nasabah {cid} risiko {risk_lv}, pengeluaran {er:.1%}. Terakhir belanja di {tx_last}. Beri 1 saran hangat personal (Bapak/Ibu) maks 3 kalimat."
+                resp = ai_client.models.generate_content(model="gemini-1.5-flash", contents=prompt)
+                msg += f"\n---\n**💡 SARAN VIRTUAL ADVISOR (GEN-AI):**\n_{resp.text}_"
+            except: pass
+            
+        return msg
+
+    def create_plots(self, u_bal, u_txn):
+        # FASE 6: VISUAL ANALYTICS
+        u_txn['m'] = u_txn['date'].dt.to_period('M').dt.to_timestamp()
+        cf = u_txn.groupby(['m', 'transaction_type'])['amount'].sum().unstack().fillna(0)
+        
+        f1 = go.Figure()
+        f1.add_trace(go.Bar(x=cf.index, y=cf.get('credit', 0), name='Pemasukan', marker_color='#82C3EB'))
+        f1.add_trace(go.Bar(x=cf.index, y=cf.get('debit', 0), name='Pengeluaran', marker_color='#0514DE'))
+        f1.update_layout(title="Laporan Arus Kas (Inflow vs Outflow)", barmode='group', template='plotly_white')
+
+        f2 = go.Figure()
+        f2.add_trace(go.Scatter(x=u_bal['month'], y=u_bal['avg_balance'], name='Saldo Rata-rata', line=dict(color='#F7BD87', width=4)))
+        f2.add_trace(go.Bar(x=u_bal['month'], y=u_bal['min_balance'], name='Saldo Minimum', marker_color='#E0EDF4', opacity=0.5))
+        f2.update_layout(title="Tren Pertumbuhan & Saldo Minimum", template='plotly_white')
+        
+        return f1, f2
 
-# --- UI ---
-engine = ArchonBankNagari()
+# --- UI LOGIC ---
+engine = ArchonNagariEngine()
 
 def run_app(cust_id):
-    data = engine.analyze(cust_id)
-    if not data: return "## ❌ ID Tidak Valid", "Gunakan C0001 - C0120", None, None
-    
-    risk_lv, score, er, u_bal, u_txn, action, reason = data
-    advice = engine.get_ai_narrative(risk_lv, er, cust_id, u_txn)
+    res = engine.analyze(cust_id)
+    if not res: return "## ❌ ID Tidak Valid", "Mohon masukkan ID C0001 - C0120", None, None
     
-    # Graphs
-    f1 = go.Figure()
-    f1.add_trace(go.Scatter(x=u_bal['month'], y=u_bal['avg_balance'], name='Avg Balance', line=dict(color='#0514DE', width=4)))
-    f1.update_layout(title="Trend Saldo (Fase 6)", template="plotly_white")
+    risk_lv, score, er, dr, u_bal, u_txn, exp, inc, action, reason, drivers = res
+    report = engine.get_summary_report(risk_lv, score, er, u_bal, exp, inc, action, reason, drivers, cust_id, u_txn)
+    p1, p2 = engine.create_plots(u_bal, u_txn)
     
-    f2 = go.Figure()
-    u_txn['m'] = u_txn['date'].dt.to_period('M').dt.to_timestamp()
-    cf = u_txn.groupby(['m', 'transaction_type'])['amount'].sum().unstack().fillna(0)
-    f2.add_trace(go.Bar(x=cf.index, y=cf.get('credit', 0), name='Inflow', marker_color='#82C3EB'))
-    f2.add_trace(go.Bar(x=cf.index, y=cf.get('debit', 0), name='Outflow', marker_color='#0514DE'))
-    f2.update_layout(title="Inflow vs Outflow", barmode='group', template='plotly_white')
-
     color = "#ef4444" if risk_lv == "HIGH" else ("#f59e0b" if risk_lv == "MEDIUM" else "#10b981")
-    report = f"""
-    ### 🛡️ HASIL AUDIT ARCHON-AI
-    - **Risk Score**: {score:.2f} ({risk_lv} RISK)
-    - **Expense Ratio**: {er:.1%}
-    
-    **🎯 REKOMENDASI NBO:**
-    **{action}** ({reason})
-    
-    **💡 SARAN VIRTUAL ADVISOR:**
-    {advice}
+    status_html = f"""
+    <div class='card-overview'>
+        <h2 style='color: #0514DE; margin:0;'>Dashboard Ringkasan</h2>
+        <div class='status-badge' style='background:{color}; margin-top:15px;'>{risk_lv} RISK LEVEL</div>
+        <p style='margin-top:10px;'><b>Risk Score:</b> {score:.2f} / 1.00</p>
+        <p><b>Expense Ratio:</b> {er:.1%}</p>
+    </div>
     """
-    
-    status_html = f"<div style='background:{color}; color:white; padding:15px; border-radius:10px; text-align:center;'><h2>STATUS: {risk_lv}</h2></div>"
-    return status_html, report, f1, f2
+    return status_html, report, p1, p2
 
 with gr.Blocks(css=custom_css) as demo:
-    gr.HTML("<div class='nagari-header'><h1>ARCHON-AI</h1></div>")
+    gr.HTML("<div class='nagari-header'><h1>🛡️ **ARCHON-AI**: BANK NAGARI</h1></div>")
+    
     with gr.Row():
         with gr.Column(scale=1):
-            id_in = gr.Textbox(label="Customer ID")
-            btn = gr.Button("ANALYZE", variant="primary")
+            id_in = gr.Textbox(label="Customer ID", placeholder="Masukkan ID (contoh: C0005)")
+            btn = gr.Button("JALANKAN ANALISIS", variant="primary")
             out_status = gr.HTML()
+        
         with gr.Column(scale=2):
             with gr.Tabs():
-                with gr.TabItem("Audit & Advice"): out_report = gr.Markdown()
-                with gr.TabItem("Visual Trends"):
-                    gr.Plot(label="Balance")
-                    gr.Plot(label="Cashflow")
+                with gr.Tab("Audit Summary"):
+                    out_report = gr.Markdown(elem_classes="report-section")
+                with gr.Tab("Visual Analytics"):
+                    plot_cf = gr.Plot(label="Cashflow Insight")
+                    plot_bal = gr.Plot(label="Balance History")
 
-    btn.click(fn=run_app, inputs=id_in, outputs=[out_status, out_report, gr.Plot(), gr.Plot()])
+    btn.click(fn=run_app, inputs=id_in, outputs=[out_status, out_report, plot_cf, plot_bal])
 
 demo.launch()