--- MODEL VALIDATION REPORT ---
Project: Digital Lending Transformation
Model: XGBoost Default Prediction Model
Version: 1.0

1. EXECUTIVE SUMMARY
   - Model Overview: XGBoost classifier for predicting loan default.
   - Performance: AUC 0.85, KS 0.42, Calibration p-value 0.18.
   - Conclusion: Model approved for production deployment.
   - Recommendations: (List any conditions or recommendations.)

2. MODEL OVERVIEW
   - Model Name: XGBoost Default Predictor
   - Model Type: XGBoost Classifier
   - Developer: Atlantic Bank Data Science Team
   - Date: 2024-06-01
   - Intended Use: Personal loan underwriting and risk assessment.

3. DATA QUALITY ASSESSMENT
   - Data Sources: Loan applications, credit bureau, transactions.
   - Sample Period: 2019-2023.
   - Data Completeness: 98.7% complete.
   - Data Quality Findings: No critical issues.

4. CONCEPTUAL SOUNDNESS
   - Methodology: XGBoost with regularisation.
   - Feature Selection: Domain expertise + automated selection.
   - Assumptions: Default patterns remain stable.
   - Limitations: May perform poorly in severe economic downturns.

5. PERFORMANCE TESTING
   - AUC: 0.85
   - KS Statistic: 0.42
   - Gini Coefficient: 0.70
   - Brier Score: 0.12
   - Calibration: Hosmer-Lemeshow p-value = 0.18

6. ROBUSTNESS TESTING
   - Out-of-Time Testing: Performance stable over 2023-2024.
   - Stress Testing: PD increases by 25% under severe stress.
   - Sensitivity Analysis: Credit score and DTI are most influential.

7. FAIRNESS TESTING
   - Disparate Impact: Ratio = 0.85 (within 4/5 rule).
   - Protected Attributes: Age, gender (proxy).
   - Recommendations: Monitor ongoing fairness.

8. VALIDATION CONCLUSION
   - Model is approved for production.
   - Conditions: (List any conditions.)
   - Next Validation Date: 2025-06-01

9. SIGN-OFFS
   - Model Developer: __________________ Date: ________
   - Model Validator: __________________ Date: ________
   - Risk Committee: __________________ Date: ________

--- API DOCUMENTATION ---
Project: Digital Lending Transformation
API: Default Prediction Service
Version: 1.0

1. OVERVIEW
   - Purpose: Provide real-time default probability predictions.
   - Base URL: https://api.atlanticbank.com/v1/predict
   - Authentication: API Key (header: X-API-Key)

2. ENDPOINTS

2.1 POST /predict
   - Description: Get default probability for a loan application.
   - Request Body (JSON):
     {
       "applicant_age": 42,
       "income": 65.4,
       "credit_score": 710,
       "dti": 28.5,
       "loan_amount": 145,
       "loan_term": 36,
       "employment_years": 8,
       "home_owner": 1,
       "marital_status": 1,
       "education": 2
     }
   - Response (JSON):
     {
       "request_id": 12345,
       "probability": 0.032,
       "prediction": 0,
       "decision": "Approved",
       "model": "XGBoost v1.0",
       "timestamp": "2024-06-01T10:00:00Z"
     }

2.2 GET /health
   - Description: Health check.
   - Response: {"status": "healthy", "version": "1.0"}

3. ERROR CODES
   - 400: Invalid request (missing fields).
   - 401: Unauthorised (invalid API key).
   - 429: Rate limit exceeded.
   - 500: Internal server error.

4. RATE LIMITS
   - 1000 requests per minute.

5. CONTACT
   - Support: datascience@atlanticbank.com

--- DEPLOYMENT GUIDE ---
Project: Digital Lending Transformation
Environment: Production (AWS)
Version: 1.0

1. DEPLOYMENT ARCHITECTURE
   - Containerised model (Docker).
   - API Gateway (Flask).
   - Orchestration (Kubernetes/EKS).
   - Monitoring (Prometheus + Grafana).
   - Logging (ELK Stack).

2. PREREQUISITES
   - AWS CLI configured.
   - kubectl installed.
   - Docker installed.
   - Access to ECR repository.

3. DEPLOYMENT STEPS
   1. Build Docker image: docker build -t default-predictor .
   2. Tag image: docker tag default-predictor:latest 123456789.dkr.ecr.us-east-1.amazonaws.com/default-predictor:latest
   3. Push image: docker push 123456789.dkr.ecr.us-east-1.amazonaws.com/default-predictor:latest
   4. Apply Kubernetes manifests: kubectl apply -f deployment.yaml
   5. Verify deployment: kubectl get pods
   6. Update service: kubectl apply -f service.yaml

4. HEALTH CHECK
   - Endpoint: GET /health
   - Expected: {"status": "healthy"}

5. ROLLBACK PROCEDURE
   - Revert to previous version: kubectl rollout undo deployment/default-predictor
   - Verify rollback: kubectl rollout status deployment/default-predictor

6. TROUBLESHOOTING
   - Check logs: kubectl logs deployment/default-predictor
   - Check pod status: kubectl describe pod <pod-name>
   - Check service: kubectl describe service default-predictor

7. CONTACT
   - MLOps Team: mlops@atlanticbank.com

--- USER GUIDE ---
Project: Digital Lending Transformation
System: AI-Powered Loan Decision Engine
Version: 1.0

1. INTRODUCTION
   Welcome to the AI-Powered Loan Decision Engine. This system helps you make faster, more accurate lending decisions using machine learning.

2. GETTING STARTED
   1. Log in with your credentials.
   2. Navigate to the Loan Application page.
   3. Enter the applicant's details.
   4. Submit for prediction.

3. USING THE SYSTEM
   3.1 Entering Applicant Details
       - Enter income, credit score, DTI, and other required fields.
       - Fields marked with * are required.

   3.2 Submitting an Application
       - Click the "Predict" button.
       - The system will return a decision within seconds.

   3.3 Understanding the Output
       - Probability: The likelihood of default (0-1).
       - Decision: Approved or Declined.
       - Explanation: Key factors influencing the decision.

4. INTERPRETING RESULTS
   - If declined, the "Explanation" section shows the top reasons.
   - You can override the decision by clicking "Override" (requires manager approval).

5. OVERRIDING A DECISION
   1. Click the "Override" button.
   2. Enter the reason for the override.
   3. Submit for manager approval.
   4. Manager reviews and approves/rejects the override.

6. REPORTING
   - View dashboards for portfolio performance and model health.
   - Generate reports for regulators and management.

7. TROUBLESHOOTING
   - If the system is slow, check your internet connection.
   - If you get an error, note the error code and contact support.

8. SUPPORT
   - Email: lending-support@atlanticbank.com
   - Phone: Ext. 1234
   - Hours: 9am-5pm, Monday-Friday

# ===================================================================
# MODULE 9, LESSON 7: CAPSTONE PROJECT – FINAL DOCUMENTATION
# ===================================================================

import pandas as pd
import numpy as np
from datetime import datetime
import matplotlib.pyplot as plt
import seaborn as sns
import warnings
warnings.filterwarnings('ignore')

print("="*70)
print("CAPSTONE PROJECT – FINAL DOCUMENTATION AND HANDOVER")
print("="*70)

# ----------------------------------------------------------------
# PART A: DOCUMENTATION CHECKLIST
# ----------------------------------------------------------------

print("\n" + "-"*60)
print("PART A: Documentation Checklist")
print("-"*60)

doc_checklist = pd.DataFrame({
    'Document': [
        'Project Charter',
        'Technical Design Document',
        'Model Development Report',
        'Model Validation Report',
        'API Documentation',
        'User Guide',
        'Deployment Guide',
        'Monitoring Guide',
        'Handover Report',
        'Lessons Learned'
    ],
    'Status': [
        '✅ Complete',
        '✅ Complete',
        '✅ Complete',
        '🟡 In Review',
        '✅ Complete',
        '🟡 In Progress',
        '✅ Complete',
        '🟡 In Progress',
        '🟡 In Progress',
        '🟡 In Progress'
    ],
    'Owner': [
        'PMO',
        'Data Engineering',
        'Data Science',
        'Validation Team',
        'MLOps',
        'Training Team',
        'MLOps',
        'MLOps',
        'Project Manager',
        'Project Team'
    ],
    'Due Date': [
        '2024-01-15',
        '2024-03-01',
        '2024-05-01',
        '2024-05-15',
        '2024-06-01',
        '2024-06-15',
        '2024-06-01',
        '2024-06-15',
        '2024-06-30',
        '2024-06-30'
    ]
})

print("Documentation Checklist:")
print(doc_checklist.to_string(index=False))

# ----------------------------------------------------------------
# PART B: VERSION CONTROL AND CHANGE LOG
# ----------------------------------------------------------------

print("\n" + "-"*60)
print("PART B: Version Control and Change Log")
print("-"*60)

change_log = pd.DataFrame({
    'Version': ['1.0', '1.1', '1.2', '2.0'],
    'Date': ['2024-05-15', '2024-05-22', '2024-06-01', '2024-06-15'],
    'Author': ['Data Science Team', 'Data Science Team', 'MLOps Team', 'Data Science Team'],
    'Changes': [
        'Initial model development and validation',
        'Fixed calibration issue; retrained model',
        'Deployment scripts and monitoring added',
        'Model retrained with new data; performance improved'
    ],
    'Approval Status': ['Approved', 'Approved', 'Approved', 'Pending']
})

print("Change Log:")
print(change_log.to_string(index=False))

# ----------------------------------------------------------------
# PART C: PROJECT HANDOVER CHECKLIST
# ----------------------------------------------------------------

print("\n" + "-"*60)
print("PART C: Project Handover Checklist")
print("-"*60)

handover_checklist = {
    "Technical Handover": [
        "Source code repositories transferred to Ops team.",
        "CI/CD pipelines configured and documented.",
        "Deployment scripts and configuration files provided.",
        "Monitoring dashboards created and accessible.",
        "API documentation completed and tested.",
        "Database schemas and data models documented.",
        "Infrastructure-as-code (Terraform/CloudFormation) provided.",
        "Security and access controls reviewed.",
        "Disaster recovery plan documented."
    ],
    "Operational Handover": [
        "Model performance baselines established.",
        "Monitoring thresholds and alerts configured.",
        "Retraining schedule and triggers documented.",
        "Support escalation paths defined.",
        "Runbooks for common issues created.",
        "Contact list for key personnel provided.",
        "System maintenance schedule established.",
        "Data backup and retention policies documented."
    ],
    "Business Handover": [
        "Training materials provided to business users.",
        "User acceptance testing (UAT) completed and signed off.",
        "Key performance indicators (KPIs) defined.",
        "Reporting dashboards created.",
        "Stakeholder communication plan finalised.",
        "Change management activities completed.",
        "Project closure presentation delivered.",
        "Lessons learned documented."
    ]
}

print("Project Handover Checklist:")
for category, items in handover_checklist.items():
    print(f"\n{category}:")
    for item in items:
        print(f"  ✅ {item}")

# ----------------------------------------------------------------
# PART D: LESSONS LEARNED
# ----------------------------------------------------------------

print("\n" + "-"*60)
print("PART D: Lessons Learned")
print("-"*60)

lessons_learned = {
    "What Went Well": [
        "Strong executive sponsorship and alignment.",
        "Excellent collaboration between data science and IT teams.",
        "Early engagement with compliance and risk teams.",
        "Iterative development with regular stakeholder demos.",
        "High-quality data due to data governance improvements.",
        "Model performance exceeded targets (AUC 0.85)."
    ],
    "Challenges and Solutions": [
        "Data quality issues → Implemented data quality framework early.",
        "Model deployment delays → Adopted MLOps and CI/CD.",
        "User resistance → Change management and training program.",
        "Regulatory uncertainty → Engaged regulators early."
    ],
    "Recommendations": [
        "Start data quality work earlier in the project.",
        "Invest in MLOps from day one.",
        "Involve end-users in design and testing.",
        "Plan for regulatory engagement from the start.",
        "Document everything – it saves time later.",
        "Celebrate successes and learn from failures."
    ]
}

print("Lessons Learned:")
for category, items in lessons_learned.items():
    print(f"\n{category}:")
    for item in items:
        print(f"  • {item}")

# ----------------------------------------------------------------
# PART E: PERFORMANCE METRICS REPORT
# ----------------------------------------------------------------

print("\n" + "-"*60)
print("PART E: Project Performance Metrics Report")
print("-"*60)

performance_metrics = pd.DataFrame({
    'Metric': [
        'Model AUC',
        'Model KS',
        'Model Calibration (H-L)',
        'Automation Rate',
        'Decision Time (Average)',
        'User Adoption Rate',
        'Customer NPS',
        'Default Rate Reduction',
        'ROI (3-year)',
        'Project Duration'
    ],
    'Target': [
        '> 0.80',
        '> 0.35',
        '> 0.05',
        '> 80%',
        '< 10 min',
        '> 80%',
        '> 60',
        '> 10%',
        '> 200%',
        '12 months'
    ],
    'Actual': [
        '0.85',
        '0.42',
        '0.18',
        '85%',
        '3.2 min',
        '92%',
        '65',
        '18%',
        '575%',
        '11.5 months'
    ],
    'Status': ['✅', '✅', '✅', '✅', '✅', '✅', '✅', '✅', '✅', '✅']
})

print("Project Performance Metrics:")
print(performance_metrics.to_string(index=False))

# ----------------------------------------------------------------
# PART F: PROJECT CLOSURE CHECKLIST
# ----------------------------------------------------------------

print("\n" + "-"*60)
print("PART F: Project Closure Checklist")
print("-"*60)

closure_checklist = [
    "✅ Project objectives achieved.",
    "✅ Success criteria met.",
    "✅ All deliverables completed.",
    "✅ Documentation complete and handed over.",
    "✅ Training completed for all users.",
    "✅ Support and maintenance agreements in place.",
    "✅ Financial closure completed (budget reconciled).",
    "✅ Project team resources released.",
    "✅ Lessons learned documented.",
    "✅ Final project report delivered.",
    "✅ Stakeholder sign-off obtained.",
    "✅ Project close-out meeting held."
]

print("Project Closure Checklist:")
for item in closure_checklist:
    print(item)

# ----------------------------------------------------------------
# PART G: PROJECT HANDOVER PACKAGE
# ----------------------------------------------------------------

print("\n" + "-"*60)
print("PART G: Project Handover Package Contents")
print("-"*60)

handover_package = {
    "Documentation": [
        "Project Charter",
        "Technical Design Document",
        "Model Development Report",
        "Model Validation Report",
        "API Documentation",
        "User Guide",
        "Deployment Guide",
        "Monitoring Guide",
        "Lessons Learned"
    ],
    "Code and Configuration": [
        "Source code (Git repository)",
        "Database schemas and migration scripts",
        "Configuration files (environments)",
        "CI/CD pipeline configurations",
        "Infrastructure-as-code (Terraform)",
        "Docker images and Dockerfiles",
        "Kubernetes manifests"
    ],
    "Data and Models": [
        "Training and validation datasets",
        "Model files (.pkl, .joblib)",
        "Feature engineering code",
        "Data quality rules and checks",
        "Monitoring data and dashboards"
    ],
    "Access and Credentials": [
        "Access to production systems",
        "API keys and credentials",
        "Database access credentials",
        "Monitoring and logging access",
        "Support and escalation contacts"
    ]
}

print("Handover Package Contents:")
for category, items in handover_package.items():
    print(f"\n{category}:")
    for item in items:
        print(f"  • {item}")

# ----------------------------------------------------------------
# PART H: SUMMARY AND RECOMMENDATIONS
# ----------------------------------------------------------------

print("\n" + "="*70)
print("PART H: Summary and Recommendations")
print("="*70)

print("""
Project Documentation and Handover – Key Takeaways:

1. Documentation is essential for regulatory compliance and operational success.
2. Different audiences require different documentation formats.
3. Model validation reports are critical for SR 11-7 compliance.
4. Technical documentation ensures smooth handover to operations.
5. User guides drive adoption and user satisfaction.
6. Lessons learned drive continuous improvement.
7. Handover packages ensure continuity after project completion.

Recommendations:
  - Start documentation early; don't leave it to the end.
  - Use templates for consistency and efficiency.
  - Review documentation with stakeholders regularly.
  - Store documentation in a central, accessible location.
  - Keep documentation up-to-date with changes.
  - Include documentation in the definition of "done."
""")

print("="*70)
print("END OF LESSON 7 – MODULE 9")
print("="*70)