update ui and use google ai

app.py

@@ -3,135 +3,141 @@ import pandas as pd
 import numpy as np
 import gradio as gr
 import plotly.graph_objects as go
-from transformers import pipeline
-from huggingface_hub import InferenceClient
+import google.generativeai as genai
 
-# --- CONFIG & STYLING ---
-HF_TOKEN = os.getenv("HF_TOKEN")
-client = InferenceClient(model="mistralai/Mistral-7B-Instruct-v0.3", token=HF_TOKEN)
+# --- KONFIGURASI GEMINI (Ganti dari Mistral ke Gemini) ---
+GOOGLE_API_KEY = os.getenv("GOOGLE_API_KEY")
+genai.configure(api_key=GOOGLE_API_KEY)
+gemini_model = genai.GenerativeModel('gemini-1.5-flash')
 
-# CSS Kustom untuk font dan tampilan mewah
+# --- CSS CUSTOM: STYLE BANK NAGARI ---
 custom_css = """
-@import url('https://fonts.googleapis.com/css2?family=Inter:wght@400;600&display=swap');
-body { font-family: 'Inter', sans-serif; background-color: #f8fafc; }
-.gradio-container { max-width: 1200px !important; }
-.risk-card { padding: 20px; border-radius: 12px; border: 1px solid #e2e8f0; background: white; }
-.high-risk { border-left: 8px solid #ef4444; }
-.medium-risk { border-left: 8px solid #f59e0b; }
-.low-risk { border-left: 8px solid #10b981; }
+@import url('https://fonts.googleapis.com/css2?family=Roboto:wght@400;700&display=swap');
+body { font-family: 'Roboto', sans-serif; background-color: #f4f4f4; }
+.gradio-container { max-width: 1250px !important; }
+.nagari-header { background-color: #800000; color: white; padding: 20px; border-radius: 10px 10px 0 0; border-bottom: 4px solid #FFD700; }
+.risk-card { padding: 25px; border-radius: 12px; background: white; box-shadow: 0 4px 6px rgba(0,0,0,0.1); border-top: 5px solid #800000; }
+.risk-high { border-left: 10px solid #d32f2f; }
+.risk-medium { border-left: 10px solid #f9a825; }
+.risk-low { border-left: 10px solid #2e7d32; }
 """
 
-class ArchonExecutive:
+class ArchonNagariEngine:
     def __init__(self):
         self.load_data()
-        try:
-            self.classifier = pipeline("text-classification", model="archon_v1")
-        except:
-            self.classifier = None
 
     def load_data(self):
-        self.df_txn = pd.read_csv('transactions.csv', parse_dates=['date'])
-        self.df_bal = pd.read_csv('balances_revised.csv', parse_dates=['month'])
-        self.df_rep = pd.read_csv('repayments_revised.csv', parse_dates=['due_date'])
+        # Fase 1: Foundation (Cek error pada data mentah)
+        self.df_txn = pd.read_csv('transactions.csv', parse_dates=['date']).fillna("")
+        self.df_cust = pd.read_csv('customers.csv').fillna(0)
+        self.df_bal = pd.read_csv('balances_revised.csv', parse_dates=['month']).fillna(0)
+        self.df_rep = pd.read_csv('repayments_revised.csv', parse_dates=['due_date']).fillna("on_time")
 
-    def calculate_logic(self, customer_id):
-        # Filter Data
+    def run_analysis(self, customer_id):
+        # Validasi Keberadaan Data
         u_txn = self.df_txn[self.df_txn['customer_id'] == customer_id].copy()
         u_bal = self.df_bal[self.df_bal['customer_id'] == customer_id].sort_values('month')
         u_rep = self.df_rep[self.df_rep['customer_id'] == customer_id]
         
-        if u_txn.empty or u_bal.empty: return None
+        if u_txn.empty or u_bal.empty:
+            return None
 
-        # --- FASE 4: DETERMINISTIC SCORING (WEIGHTED) ---
+        # --- FASE 4: RISK SCORING (WEIGHTED 30/20/20/20/10) ---
         income = u_txn[u_txn['transaction_type'] == 'credit']['amount'].sum()
         expense = u_txn[u_txn['transaction_type'] == 'debit']['amount'].sum()
         er = expense / income if income > 0 else 1.0
         
-        # Scoring per variabel (0, 0.5, 1)
+        # Scoring Rules (Fase 4)
         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
+        vol_s = 0.5 # Placeholder Volatility
         
-        # Rumus Bobot Manajemen (Fase 4)
-        score = (0.3 * er_s) + (0.2 * bt_s) + (0.2 * od_s) + (0.2 * mp_s) + 0.1
-        risk_lv = "HIGH" if score >= 0.7 else ("MEDIUM" if score >= 0.4 else "LOW")
+        final_score = (0.3 * er_s) + (0.2 * bt_s) + (0.2 * od_s) + (0.2 * mp_s) + (0.1 * vol_s)
+        risk_lv = "HIGH" if final_score >= 0.7 else ("MEDIUM" if final_score >= 0.4 else "LOW")
         
-        return risk_lv, score, er, u_bal, u_txn
+        return risk_lv, final_score, er, u_bal, u_txn
 
-    def get_llm_advice(self, risk_lv, er, cust_id, u_txn):
-        # FASE 5: NBO ENGINE (LLM)
-        last_txn = u_txn.iloc[-1]['raw_description'] if not u_txn.empty else "N/A"
-        
-        prompt = f"<s>[INST] Anda adalah AI Financial Advisor Bank. Nasabah {cust_id} memiliki risiko {risk_lv} dengan pengeluaran {er:.1%}. Transaksi terakhirnya adalah '{last_txn}'. Berikan 1 saran finansial singkat yang SANGAT NATURAL (tidak kaku), empati, dan solutif dalam Bahasa Indonesia. Jangan gunakan kata 'Nasabah'. [/INST]</s>"
+    def get_gemini_advice(self, risk_lv, er, cust_id, u_txn):
+        # FASE 5: NBO DENGAN GEMINI (Sangat Natural & Tidak Template)
+        tx_history = u_txn.tail(5)['raw_description'].tolist()
+        context = f"Nasabah {cust_id} memiliki risiko {risk_lv}, rasio belanja {er:.1%}. Transaksi terakhir: {', '.join(tx_history)}."
         
+        prompt = f"""
+        Tugas: Sebagai Virtual Advisor Bank Nagari yang bijak.
+        Data: {context}
+        Instruksi: Berikan saran finansial yang personal, hangat, dan solutif dalam Bahasa Indonesia. 
+        Gunakan sapaan 'Bapak/Ibu'. Jangan terlihat seperti bot. Fokus pada peningkatan ketahanan finansial (Resilience).
+        Maksimal 3 kalimat.
+        """
         try:
-            response = client.text_generation(prompt, max_new_tokens=150, temperature=0.7)
-            return response.strip()
+            response = gemini_model.generate_content(prompt)
+            return response.text
         except:
-            return "Halo! Kami melihat rasio pengeluaran Anda cukup tinggi. Disarankan untuk mulai memantau pengeluaran discretionary bulan ini."
+            return "Kami menyarankan Bapak/Ibu untuk meninjau kembali pos pengeluaran bulan ini agar saldo tetap terjaga sehat."
 
     def create_viz(self, u_bal, u_txn):
-        # Grafik Tren Saldo
+        # FASE 6: VISUALIZATION (BANK NAGARI STYLE)
+        # 1. Trend Saldo
         fig_bal = go.Figure()
-        fig_bal.add_trace(go.Scatter(x=u_bal['month'], y=u_bal['avg_balance'], name='Avg Balance', line=dict(color='#10b981', width=4)))
-        fig_bal.add_trace(go.Bar(x=u_bal['month'], y=u_bal['min_balance'], name='Min Balance', marker_color='#94a3b8', opacity=0.4))
-        fig_bal.update_layout(title="Kesehatan Saldo (Fase 6)", template="plotly_white", margin=dict(t=40, b=0, l=0, r=0))
+        fig_bal.add_trace(go.Scatter(x=u_bal['month'], y=u_bal['avg_balance'], name='Rata-rata Saldo', line=dict(color='#800000', width=4)))
+        fig_bal.add_trace(go.Bar(x=u_bal['month'], y=u_bal['min_balance'], name='Saldo Minimum', marker_color='#FFD700', opacity=0.5))
+        fig_bal.update_layout(title="Laporan Tren Saldo Bulanan", template="plotly_white", legend=dict(orientation="h"))
 
-        # Grafik Income vs Expense
-        u_txn['month'] = u_txn['date'].dt.to_period('M').dt.to_timestamp()
-        monthly = u_txn.groupby(['month', 'transaction_type'])['amount'].sum().unstack().fillna(0)
-        
+        # 2. Income vs Expense
+        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)
         fig_cf = go.Figure()
-        fig_cf.add_trace(go.Bar(x=monthly.index, y=monthly.get('credit', 0), name='Income', marker_color='#3b82f6'))
-        fig_cf.add_trace(go.Bar(x=monthly.index, y=monthly.get('debit', 0), name='Expense', marker_color='#f43f5e'))
-        fig_cf.update_layout(title="Income vs Expense", barmode='group', template="plotly_white", margin=dict(t=40, b=0, l=0, r=0))
+        fig_cf.add_trace(go.Bar(x=cf.index, y=cf.get('credit', 0), name='Pemasukan', marker_color='#2e7d32'))
+        fig_cf.add_trace(go.Bar(x=cf.index, y=cf.get('debit', 0), name='Pengeluaran', marker_color='#800000'))
+        fig_cf.update_layout(title="Arus Kas Pemasukan vs Pengeluaran", barmode='group', template="plotly_white")
         
         return fig_bal, fig_cf
 
 # --- INTERFACE ---
-engine = ArchonExecutive()
+engine = ArchonNagariEngine()
 
-def analyze_and_show(cust_id):
-    data = engine.calculate_logic(cust_id)
-    if not data: return "## ❌ ID Tidak Ditemukan", "Mohon masukkan ID C0001 - C0120", None, None
+def run_archon(cust_id):
+    res = engine.run_analysis(cust_id)
+    if not res:
+        return "### ⚠️ Data Tidak Ditemukan", "ID tidak terdaftar di sistem.", None, None
     
-    risk_lv, score, er, u_bal, u_txn = data
-    advice = engine.get_llm_advice(risk_lv, er, cust_id, u_txn)
-    v_bal, v_cf = engine.create_viz(u_bal, u_txn)
+    risk_lv, score, er, u_bal, u_txn = res
+    advice = engine.get_gemini_advice(risk_lv, er, cust_id, u_txn)
+    p_bal, p_cf = engine.create_viz(u_bal, u_txn)
     
-    color_class = "high-risk" if risk_lv == "HIGH" else ("medium-risk" if risk_lv == "MEDIUM" else "low-risk")
+    status_cls = "risk-high" if risk_lv == "HIGH" else ("risk-medium" if risk_lv == "MEDIUM" else "risk-low")
     
-    report_md = f"""
-    <div class="risk-card {color_class}">
-        <h3>🛡️ Hasil Analisis: {risk_lv}</h3>
+    report_html = f"""
+    <div class="risk-card {status_cls}">
+        <h2 style="color: #800000; margin-top:0;">📋 Ringkasan Risiko: {risk_lv}</h2>
         <p><b>Risk Score:</b> {score:.2f} | <b>Expense Ratio:</b> {er:.1%}</p>
-        <hr>
-        <p><i>Sistem mendeteksi aktivitas keuangan nasabah memerlukan perhatian khusus pada manajemen saldo harian.</i></p>
+        <p style="font-size: 0.9em; color: #666;">Berdasarkan bobot parameter Fase 4: Pengeluaran, Saldo, dan Riwayat Cicilan.</p>
     </div>
     """
-    return report_md, advice, v_bal, v_cf
+    return report_html, advice, p_bal, p_cf
 
-with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
-    gr.Markdown("# 🏦 ARCHON: Financial Intelligence Dashboard")
-    gr.Markdown("Transformasi data transaksi menjadi insight prediktif dan aksi nyata.")
-    
-    with gr.Row():
+with gr.Blocks(css=custom_css) as demo:
+    with gr.Div(elem_classes="nagari-header"):
+        gr.Markdown("# 🛡️ ARCHON-AI: FINANCIAL RESILIENCE ENGINE")
+        gr.Markdown("Pusat Intelijen Manajemen Risiko & Perilaku Nasabah")
+
+    with gr.Row(variant="panel"):
         with gr.Column(scale=1):
-            input_id = gr.Textbox(label="Customer ID", placeholder="e.g. C0005")
-            btn = gr.Button("RUN ANALYSIS", variant="primary")
-            report_out = gr.HTML()
+            id_input = gr.Textbox(label="Customer ID", placeholder="e.g., C0014")
+            btn = gr.Button("PROSES ANALISIS", variant="primary")
+            out_report = gr.HTML()
         
         with gr.Column(scale=2):
             with gr.Tabs():
-                with gr.TabItem("Balance Trend"):
-                    plot_bal = gr.Plot()
-                with gr.TabItem("Cashflow Insight"):
+                with gr.TabItem("Analisis Arus Kas"):
                     plot_cf = gr.Plot()
-                    
-    with gr.Row():
-        advice_out = gr.Textbox(label="Archon AI Advice (Pilar 3 & 5)", lines=4)
+                with gr.TabItem("Tren Saldo"):
+                    plot_bal = gr.Plot()
+
+    out_advice = gr.Textbox(label="Saran AI Virtual Advisor (Powered by Gemini)", lines=4)
 
-    btn.click(fn=analyze_and_show, inputs=input_id, outputs=[report_out, advice_out, plot_bal, plot_cf])
+    btn.click(fn=run_archon, inputs=id_input, outputs=[out_report, out_advice, plot_bal, plot_cf])
 
 demo.launch()