Upload 6 files

advisor_bot.py

@@ -0,0 +1,46 @@
+from langchain_groq import ChatGroq
+from langchain_core.prompts import PromptTemplate
+import os
+from dotenv import load_dotenv
+load_dotenv()
+
+llm = ChatGroq(model="llama-3.1-8b-instant", api_key=os.getenv("GROQ_API_KEY"))
+
+prompt = PromptTemplate.from_template("""
+You are RiskGuard AI, a professional digital bank assistant.
+
+Below is the user's credit evaluation:
+
+• Probability of Default: {probability}%
+• Credit Score: {credit_score}
+• Rating Category: {rating}
+
+Write a short, friendly message (4–6 lines) following this format:
+
+1) A polite greeting such as:
+   "Thank you for using RiskGuard AI for your loan assessment."
+
+2) A clear decision tone:
+   - If the score and risk level are strong: indicate the loan is likely suitable for approval.
+   - If risk is high: indicate that approval may be difficult at this time.
+
+3) Give one or two simple, actionable suggestions for improvement (if needed).
+
+4) Close with a short support line such as:
+   "If you have questions or want guidance, feel free to talk to our loan advisor chatbot."
+
+Tone: concise, professional, supportive. No long bullet lists, no emojis, no legal claims.
+""")
+
+
+def generate_advice(probability, credit_score, rating):
+    formatted_prompt = prompt.format(
+        probability=round(probability * 100, 2),
+        credit_score=credit_score,
+        rating=rating
+    )
+
+    result = llm.invoke(formatted_prompt)
+    return result.content
+
+

artifacts/model_data.joblib

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e127bcce15dacbe33e11988528e5219f89a1ef04f19662ab688d334a7fe90c49
+size 4769

chatbot_advisor.py

@@ -0,0 +1,97 @@
+from langgraph.graph import StateGraph, START, END
+from typing import TypedDict,Annotated
+from langchain_core.messages import HumanMessage, BaseMessage, AIMessage, SystemMessage
+from langchain_openai import ChatOpenAI
+from langgraph.checkpoint.memory import MemorySaver
+from dotenv import load_dotenv
+from langgraph.graph.message import add_messages
+load_dotenv()
+
+class ChatState(TypedDict):
+    messages : Annotated[list[BaseMessage], add_messages]
+
+llm = ChatOpenAI(model='gpt-4.1-nano', streaming=True)
+
+SYSTEM_MESSAGE = SystemMessage(
+    content=(
+        "You are RiskGuard AI, an intelligent credit risk and financial guidance assistant. "
+        "Your role is to help users understand their credit standing, loan eligibility, and financial risk profile "
+        "based on the information provided to you. "
+        
+        "Your responses should be:\n"
+        "- Clear, concise, and easy to understand (avoid technical jargon unless needed)\n"
+        "- Professional and non-judgmental in tone\n"
+        "- Supportive, encouraging, and solution-focused\n"
+        "- Insightful, offering actionable steps the user can take to improve\n"
+        "- Aligned with responsible financial communication (no promises, guarantees, or legal statements)\n\n"
+        
+        "When answering:\n"
+        "- Reference relevant financial data if provided\n"
+        "- Offer practical recommendations that feel personalized\n"
+        "- Keep responses conversational, modern, and human-like, similar to a digital bank assistant or financial coach\n\n"
+        
+        "If the user asks about next steps, provide helpful financial strategies such as improving repayment history, "
+        "reducing utilization, maintaining fewer inquiries, or improving documentation.\n\n"
+        
+        "If you do not have enough information to answer accurately, ask a clarifying question.\n\n"
+        
+        "Never provide legal, tax, or investment guarantees.\n\n"
+        
+        "Your priority is to help the user feel informed, supported, and confident in managing their credit journey."
+    )
+)
+
+
+def chat_node(state : ChatState):
+    user_query = state['messages']
+    query = [SYSTEM_MESSAGE]+user_query
+    response = llm.invoke(query)
+    return {'messages': [response]}
+
+checkpointer = MemorySaver()
+graph = StateGraph(ChatState)
+graph.add_node("chat_node", chat_node)
+graph.add_edge(START, 'chat_node')
+graph.add_edge('chat_node', END)
+
+financial_advisor_chatbot = graph.compile(checkpointer=checkpointer)
+
+thread_id='1'
+config = {'configurable' : {'thread_id' : thread_id}}
+
+def format_chat_input(probability, credit_score, rating, advisor_reply, user_message):
+    return f"""
+Below is the most recent loan evaluation details. Use them when responding.
+
+CREDIT ANALYSIS
+---------------
+• Credit Score: {credit_score}
+• Default Probability: {probability:.2%}
+• Rating: {rating}
+
+AI ADVISOR SUMMARY
+------------------
+{advisor_reply}
+
+USER QUESTION
+-------------
+{user_message}
+
+Respond as RiskGuard AI in a clear, concise and helpful tone.
+"""
+
+def ask_chatbot(probability, credit_score, rating, advisor_reply, user_message, thread_id):
+    formatted_msg = format_chat_input(probability, credit_score, rating, advisor_reply, user_message)
+
+    initial_state = {
+        "messages": [HumanMessage(content=formatted_msg)]
+    }
+
+    config = {"configurable": {"thread_id": thread_id}}
+
+    response = financial_advisor_chatbot.invoke(initial_state, config=config)
+    return response["messages"][-1].content
+
+
+
+

main.py

@@ -0,0 +1,456 @@
+import streamlit as st
+import requests
+import uuid
+from datetime import datetime
+
+# Page configuration
+st.set_page_config(
+    page_title="RiskGuard AI: Credit Risk Modelling",
+    page_icon="🛡️",
+    layout="wide"
+)
+
+# Initialize session storage
+if "chat_history" not in st.session_state:
+    st.session_state.chat_history = []
+if "thread_id" not in st.session_state:
+    st.session_state.thread_id = str(uuid.uuid4())
+if "analysis_done" not in st.session_state:
+    st.session_state.analysis_done = False
+if "show_info" not in st.session_state:
+    st.session_state.show_info = False
+
+# ========================= ENHANCED UI STYLING (UPDATED) =========================
+st.markdown("""
+<style>
+@import url('https://fonts.googleapis.com/css2?family=Montserrat:wght@400;600;700&display=swap');
+* {
+    font-family: 'Montserrat', 'Inter', sans-serif;
+}
+.main {
+    background: linear-gradient(135deg, #212d3b 0%, #223a57 100%);
+    padding: 2rem 0;
+}
+.header-container {
+    background: linear-gradient(135deg, #243b55 0%, #141e30 100%);
+    padding: 2rem;
+    border-radius: 16px;
+    box-shadow: 0 8px 32px rgba(36, 59, 85,0.13);
+    margin-bottom: 2rem;
+    text-align: center;
+}
+.header-title {
+    color: white;
+    font-size: 2.5rem;
+    font-weight: 700;
+    margin: 0;
+    text-shadow: 2px 2px 6px rgba(20,30,48,0.2);
+    font-family: 'Montserrat', sans-serif;
+}
+.header-subtitle {
+    color: #aab3cf;
+    font-size: 1.1rem;
+    margin-top: 0.5rem;
+    font-family: 'Montserrat', sans-serif;
+}
+.section-card {
+    background: #25304b;
+    padding: 1.5rem;
+    border-radius: 12px;
+    box-shadow: 0 4px 8px rgba(36, 59, 85,0.12);
+    margin-bottom: 1.5rem;
+    border-left: 4px solid #2952a3;
+    color: #f3f4fa;
+}
+.section-title {
+    color: #f1f7fc;
+    font-size: 1.3rem;
+    font-weight: 700;
+    margin-bottom: 1rem;
+    display: flex;
+    align-items: center;
+    gap: 0.5rem;
+    font-family: 'Montserrat', sans-serif;
+}
+.info-box {
+    background: linear-gradient(135deg, #344667 0%, #2952a3 100%);
+    color: #f1f7fc;
+    padding: 1rem;
+    border-radius: 10px;
+    margin: 1rem 0;
+    border-left: 4px solid #18aad5;
+    animation: slideIn 0.5s ease-out;
+}
+@keyframes slideIn {
+    from { opacity: 0; transform: translateY(-10px); }
+    to { opacity: 1; transform: translateY(0); }
+}
+.metric-card {
+    background: linear-gradient(135deg, #2952a3 0%, #243b55 100%);
+    padding: 1.2rem;
+    border-radius: 12px;
+    color: #f3f4fa;
+    text-align: center;
+    box-shadow: 0 4px 15px rgba(41, 82, 163, 0.2);
+    transition: transform 0.3s ease;
+}
+.metric-card:hover {
+    transform: translateY(-5px);
+}
+.metric-value {
+    font-size: 2rem;
+    font-weight: 700;
+    margin: 0.5rem 0;
+    font-family: 'Montserrat', sans-serif;
+}
+.metric-label {
+    font-size: 0.9rem;
+    opacity: 0.92;
+}
+.result-card {
+    background: linear-gradient(135deg, #333b9b 0%, #2d375b 100%);
+    padding: 2rem;
+    border-radius: 16px;
+    color: #f4f8ff;
+    box-shadow: 0 8px 32px rgba(51, 59, 155, 0.10);
+    margin: 1.5rem 0;
+}
+.result-grid {
+    display: grid;
+    grid-template-columns: repeat(auto-fit, minmax(200px, 1fr));
+    gap: 1.5rem;
+    margin-top: 1rem;
+}
+.result-item {
+    background: rgba(25,40,65,0.2);
+    padding: 1.5rem;
+    border-radius: 12px;
+    backdrop-filter: blur(10px);
+}
+.result-item-label {
+    font-size: 0.9rem;
+    opacity: 0.9;
+    margin-bottom: 0.5rem;
+}
+.result-item-value {
+    font-size: 1.8rem;
+    font-weight: 700;
+}
+.advisor-box {
+    background: linear-gradient(135deg, #2e8bcb 0%, #243b55 100%);
+    color: #f5f5fa;
+    padding: 1.5rem;
+    border-radius: 12px;
+    margin-top: 1.5rem;
+    border-left: 4px solid #18aad5;
+    box-shadow: 0 4px 15px rgba(46, 139, 203, 0.1);
+}
+.advisor-box h4 {
+    color: #f1f7fc;
+    margin-top: 0;
+}
+.chat-container {
+    background: #24304e;
+    color: #fff;
+    padding: 1.5rem;
+    border-radius: 12px;
+    box-shadow: 0 4px 6px rgba(36, 48, 78,0.12);
+    max-height: 400px;
+    overflow-y: auto;
+    margin-bottom: 1rem;
+}
+.chat-bubble-user {
+    background: linear-gradient(135deg, #2952a3 0%, #243b55 100%);
+    color: white;
+    padding: 12px 18px;
+    border-radius: 18px 18px 4px 18px;
+    margin: 8px 0 8px auto;
+    max-width: 70%;
+    text-align: right;
+    box-shadow: 0 2px 8px rgba(41,82,163, 0.13);
+    animation: slideInRight 0.3s ease-out;
+    font-family: 'Montserrat', sans-serif;
+}
+.chat-bubble-bot {
+    background: #20293b;
+    border: 2px solid #2a3d6a;
+    color: #ebf2f8;
+    padding: 12px 18px;
+    border-radius: 18px 18px 18px 4px;
+    margin: 8px auto 8px 0;
+    max-width: 70%;
+    box-shadow: 0 2px 8px rgba(0,0,0,0.1);
+    animation: slideInLeft 0.3s ease-out;
+    font-family: 'Montserrat', sans-serif;
+}
+@keyframes slideInRight {
+    from { opacity: 0; transform: translateX(20px); }
+    to { opacity: 1; transform: translateX(0); }
+}
+@keyframes slideInLeft {
+    from { opacity: 0; transform: translateX(-20px); }
+    to { opacity: 1; transform: translateX(0); }
+}
+.stButton>button {
+    width: 100%;
+    background: linear-gradient(135deg, #285aeb 0%, #223a57 100%);
+    color: white;
+    font-weight: 600;
+    border: none;
+    border-radius: 10px;
+    padding: 0.8rem;
+    font-size: 1rem;
+    font-family: 'Montserrat', sans-serif;
+    transition: all 0.3s ease;
+    box-shadow: 0 4px 15px rgba(40, 90, 235, 0.13);
+}
+.stButton>button:hover {
+    transform: translateY(-2px);
+    box-shadow: 0 6px 20px rgba(40, 90, 235, 0.20);
+}
+.progress-container {
+    margin: 1rem 0;
+}
+.stNumberInput>div>div>input,
+.stSelectbox>div>div>select,
+.stTextInput>div>div>input {
+    border-radius: 8px;
+    border: 2px solid #233269;
+    padding: 0.5rem;
+    transition: border-color 0.3s ease;
+    background: #2d375b;
+    color: #e9ecfa;
+    font-family: 'Montserrat', sans-serif;
+}
+.stNumberInput>div>div>input:focus,
+.stSelectbox>div>div>select:focus,
+.stTextInput>div>div>input:focus {
+    border-color: #285aeb;
+    box-shadow: 0 0 0 3px rgba(40, 90, 235, 0.1);
+}
+.alert-banner {
+    background: linear-gradient(135deg, #233269 0%, #285aeb 100%);
+    color: #fff;
+    padding: 1rem;
+    border-radius: 10px;
+    margin: 1rem 0;
+    border-left: 4px solid #18aad5;
+    display: flex;
+    align-items: center;
+    gap: 0.5rem;
+}
+.chat-container::-webkit-scrollbar {
+    width: 8px;
+}
+.chat-container::-webkit-scrollbar-track {
+    background: #1a2336;
+    border-radius: 10px;
+}
+.chat-container::-webkit-scrollbar-thumb {
+    background: #2952a3;
+    border-radius: 10px;
+}
+.chat-container::-webkit-scrollbar-thumb:hover {
+    background: #223a57;
+}
+</style>
+""", unsafe_allow_html=True)
+
+# ========================= HEADER =========================
+st.markdown("""
+<div class="header-container">
+    <h1 class="header-title">🛡️ RiskGuard AI</h1>
+    <p class="header-subtitle">Advanced AI-Powered Credit Risk Assessment Platform</p>
+</div>
+""", unsafe_allow_html=True)
+
+# Alert Banner
+st.markdown("""
+<div class="alert-banner">
+    ⚠️ <strong>Note:</strong> First request may take up to 20 seconds (API cold start).
+</div>
+""", unsafe_allow_html=True)
+
+# Info Toggle
+if st.button("ℹ️ How It Works"):
+    st.session_state.show_info = not st.session_state.show_info
+
+if st.session_state.show_info:
+    st.markdown("""
+    <div class="info-box">
+        <h4>📊 About RiskGuard AI</h4>
+        <p><strong>What we analyze:</strong></p>
+        <ul>
+            <li>Personal financial profile and credit history</li>
+            <li>Loan-to-income ratio and debt burden</li>
+            <li>Payment patterns and delinquency records</li>
+            <li>Credit utilization and account management</li>
+        </ul>
+        <p><strong>Our AI provides:</strong></p>
+        <ul>
+            <li>Default probability predictions</li>
+            <li>Credit score assessment</li>
+            <li>Risk rating classification</li>
+            <li>Personalized recommendations</li>
+        </ul>
+    </div>
+    """, unsafe_allow_html=True)
+
+# ========================= INPUT FORM =========================
+col_left, col_right = st.columns([1, 1], gap="large")
+
+with col_left:
+    st.markdown('<div class="section-card">', unsafe_allow_html=True)
+    st.markdown('<div class="section-title">👤 Personal & Loan Information</div>', unsafe_allow_html=True)
+    
+    age = st.number_input("Age", 18, 100, 28, help="Applicant's age in years")
+    income = st.number_input("Annual Income (₹)", 0, 50000000, 1200000, step=50000, help="Total annual income")
+    loan_amount = st.number_input("Loan Amount (₹)", 0, 50000000, 2500000, step=100000, help="Requested loan amount")
+    loan_tenure_months = st.number_input("Loan Tenure (months)", 0, 480, 36, help="Loan repayment period")
+    
+    loan_purpose = st.selectbox("Loan Purpose", ["Education", "Home", "Auto", "Personal"])
+    residence_type = st.selectbox("Residence Type", ["Owned", "Rented", "Mortgage"])
+    loan_type = st.selectbox("Loan Type", ["Secured", "Unsecured"])
+    
+    st.markdown('</div>', unsafe_allow_html=True)
+    
+    loan_to_income_ratio = loan_amount / income if income > 0 else 0
+    st.markdown(f"""
+    <div class="metric-card">
+        <div class="metric-label">Loan-to-Income Ratio</div>
+        <div class="metric-value">{loan_to_income_ratio:.2f}x</div>
+    </div>
+    """, unsafe_allow_html=True)
+
+with col_right:
+    st.markdown('<div class="section-card">', unsafe_allow_html=True)
+    st.markdown('<div class="section-title">💳 Credit Profile</div>', unsafe_allow_html=True)
+    avg_dpd_per_delinquency = st.number_input("Average Days Past Due", 0, 200, 20, help="Average days past due per delinquency")
+    delinquency_ratio = st.number_input("Delinquency Ratio (%)", 0, 100, 30, help="Percentage of delinquent accounts")
+    credit_utilization_ratio = st.number_input("Credit Utilization (%)", 0, 100, 30, help="Percentage of available credit used")
+    num_open_accounts = st.number_input("Open Loan Accounts", 0, 20, 2, help="Number of currently active loan accounts")
+    st.markdown('</div>', unsafe_allow_html=True)
+
+# ========================= ANALYSIS BUTTON =========================
+st.markdown("<br>", unsafe_allow_html=True)
+
+if st.button("🔍 Analyze Credit Risk", use_container_width=True):
+    API_URL = st.secrets["API_URL"]
+
+    payload = {
+        "age": age,
+        "income": income,
+        "loan_amount": loan_amount,
+        "loan_tenure_months": loan_tenure_months,
+        "avg_dpd_per_delinquency": avg_dpd_per_delinquency,
+        "delinquency_ratio": delinquency_ratio,
+        "credit_utilization_ratio": credit_utilization_ratio,
+        "num_open_accounts": num_open_accounts,
+        "residence_type": residence_type,
+        "loan_purpose": loan_purpose,
+        "loan_type": loan_type
+    }
+
+    with st.spinner("🤖 Analyzing your credit profile..."):
+        try:
+            r = requests.post(API_URL, json=payload, timeout=30)
+            if r.status_code == 200:
+                result = r.json()
+
+                # Display Results
+                st.markdown('<div class="result-card">', unsafe_allow_html=True)
+                st.markdown('<h3 style="margin-top:0; color:white;">📊 Assessment Results</h3>', unsafe_allow_html=True)
+                st.markdown(f"""
+                <div class="result-grid">
+                    <div class="result-item">
+                        <div class="result-item-label">Default Probability</div>
+                        <div class="result-item-value">{result['probability']:.2%}</div>
+                    </div>
+                    <div class="result-item">
+                        <div class="result-item-label">Credit Score</div>
+                        <div class="result-item-value">{result['credit_score']}</div>
+                    </div>
+                    <div class="result-item">
+                        <div class="result-item-label">Risk Rating</div>
+                        <div class="result-item-value">{result['rating']}</div>
+                    </div>
+                </div>
+                """, unsafe_allow_html=True)
+                st.markdown('</div>', unsafe_allow_html=True)
+
+                if result.get("advisor_response"):
+                    st.markdown(f"""
+                    <div class="advisor-box">
+                        <h4>🤖 AI Credit Advisor Insights</h4>
+                        <p>{result['advisor_response']}</p>
+                    </div>
+                    """, unsafe_allow_html=True)
+
+                st.session_state.probability = result["probability"]
+                st.session_state.credit_score = result["credit_score"]
+                st.session_state.rating = result["rating"]
+                st.session_state.advisor_reply = result["advisor_response"]
+                st.session_state.analysis_done = True
+                
+                st.success("✅ Analysis complete! You can now chat with our AI assistant below.")
+
+            else:
+                st.error(f"❌ API Error: {r.status_code} - {r.text}")
+        except requests.exceptions.Timeout:
+            st.error("⏱️ Request timed out. Please try again.")
+        except Exception as e:
+            st.error(f"❌ Request failed: {e}")
+
+# ========================= CHATBOT =========================
+if st.session_state.analysis_done:
+    st.markdown("<br><br>", unsafe_allow_html=True)
+    st.markdown('<div class="section-card">', unsafe_allow_html=True)
+    st.markdown('<div class="section-title">💬 Interactive Loan Chat Assistant</div>', unsafe_allow_html=True)
+    if st.session_state.chat_history:
+        st.markdown('<div class="chat-container">', unsafe_allow_html=True)
+        for role, msg in st.session_state.chat_history:
+            bubble = "chat-bubble-user" if role == "user" else "chat-bubble-bot"
+            prefix = "You: " if role == "user" else "🤖 Assistant: "
+            st.markdown(f"<div class='{bubble}'><strong>{prefix}</strong>{msg}</div>", unsafe_allow_html=True)
+        st.markdown('</div>', unsafe_allow_html=True)
+    user_query = st.text_input("Ask a question about your credit assessment:", placeholder="e.g., How can I improve my credit score?")
+    col_send, col_clear = st.columns([3, 1])
+    with col_send:
+        send_button = st.button("📤 Send Message", use_container_width=True)
+    with col_clear:
+        if st.button("🗑️ Clear Chat", use_container_width=True):
+            st.session_state.chat_history = []
+            st.session_state.thread_id = str(uuid.uuid4())
+            st.experimental_rerun()
+    if send_button and user_query.strip():
+        CHAT_URL = st.secrets["CHAT_URL"]
+        payload = {
+            "thread_id": st.session_state.thread_id,
+            "message": user_query,
+            "probability": st.session_state.probability,
+            "credit_score": st.session_state.credit_score,
+            "rating": st.session_state.rating,
+            "advisor_reply": st.session_state.advisor_reply
+        }
+        with st.spinner("🤖 Thinking..."):
+            try:
+                r = requests.post(CHAT_URL, json=payload, timeout=30)
+                if r.status_code == 200:
+                    reply = r.json()["response"]
+                    st.session_state.chat_history.append(("user", user_query))
+                    st.session_state.chat_history.append(("bot", reply))
+                    st.experimental_rerun()
+                else:
+                    st.error(f"❌ Chat server error: {r.status_code}")
+            except Exception as e:
+                st.error(f"❌ Chat failed: {e}")
+    st.markdown('</div>', unsafe_allow_html=True)
+
+# Footer
+st.markdown("<br><br>", unsafe_allow_html=True)
+st.markdown("""
+<div style='text-align: center; color: #7f8c8d; font-size: 0.9rem;'>
+    <p>🛡️ RiskGuard AI © 2025 | Powered by Advanced Machine Learning</p>
+    <p style='font-size: 0.8rem;'>For demonstration purposes only. Not financial advice.</p>
+</div>
+""", unsafe_allow_html=True)

prediction_helper.py

@@ -0,0 +1,80 @@
+import joblib
+import numpy as np
+import pandas as pd
+from sklearn.preprocessing import MinMaxScaler
+
+MODEL_PATH = 'artifacts/model_data.joblib'
+
+model_data = joblib.load(MODEL_PATH)
+model = model_data['model']
+scaler = model_data['scaler']
+features = model_data['features']
+cols_to_scale = model_data['cols_to_scale']
+
+def prepare_input(age, income, loan_amount, loan_tenure_months, avg_dpd_per_delinquency,
+                    delinquency_ratio, credit_utilization_ratio, num_open_accounts, residence_type,
+                    loan_purpose, loan_type):
+    input_data = {
+        'age': age,
+        'loan_tenure_months': loan_tenure_months,
+        'number_of_open_accounts': num_open_accounts,
+        'credit_utilization_ratio': credit_utilization_ratio,
+        'loan_to_income': loan_amount / income if income > 0 else 0,
+        'delinquency_ratio': delinquency_ratio,
+        'avg_dpd_per_delinquency': avg_dpd_per_delinquency,
+        'residence_type_Owned': 1 if residence_type == 'Owned' else 0,
+        'residence_type_Rented': 1 if residence_type == 'Rented' else 0,
+        'loan_purpose_Education': 1 if loan_purpose == 'Education' else 0,
+        'loan_purpose_Home': 1 if loan_purpose == 'Home' else 0,
+        'loan_purpose_Personal': 1 if loan_purpose == 'Personal' else 0,
+        'loan_type_Unsecured': 1 if loan_type == 'Unsecured' else 0,
+        'number_of_dependants': 1, 
+        'years_at_current_address': 1,  
+        'zipcode': 1,
+        'sanction_amount': 1,
+        'processing_fee': 1, 
+        'gst': 1, 
+        'net_disbursement': 1,
+        'principal_outstanding': 1,
+        'bank_balance_at_application': 1,
+        'number_of_closed_accounts': 1, 
+        'enquiry_count': 1
+    }
+
+    df = pd.DataFrame([input_data])
+    df[cols_to_scale] = scaler.transform(df[cols_to_scale])
+    df = df[features]
+    return df
+
+def predict(age, income, loan_amount, loan_tenure_months, avg_dpd_per_delinquency,
+            delinquency_ratio, credit_utilization_ratio, num_open_accounts,
+            residence_type, loan_purpose, loan_type):
+    input_df = prepare_input(age, income, loan_amount, loan_tenure_months, avg_dpd_per_delinquency,
+                             delinquency_ratio, credit_utilization_ratio, num_open_accounts, residence_type,
+                             loan_purpose, loan_type)
+
+    probability, credit_score, rating = calculate_credit_score(input_df)
+
+    return probability, credit_score, rating
+
+def calculate_credit_score(input_df, base_score=300, scale_length=600):
+    x = np.dot(input_df.values, model.coef_.T) + model.intercept_
+
+    default_probability = 1 / (1 + np.exp(-x))
+    non_default_probability = 1 - default_probability
+    credit_score = base_score + non_default_probability.flatten() * scale_length
+    def get_rating(score):
+        if 300 <= score < 500:
+            return 'Poor'
+        elif 500 <= score < 650:
+            return 'Average'
+        elif 650 <= score < 750:
+            return 'Good'
+        elif 750 <= score <= 900:
+            return 'Excellent'
+        else:
+            return 'Undefined'
+    rating = get_rating(credit_score[0])
+    return default_probability.flatten()[0], int(credit_score[0]), rating
+
+