📊 ICONNET Predictive Analytics Dashboard

🎯 Overview

Dashboard analisis prediktif untuk optimalisasi pendapatan layanan ICONNET di PT PLN ICON PLUS SBU Regional Jawa Tengah. Dashboard ini menggunakan machine learning untuk prediksi churn dan segmentasi pelanggan.

✨ Features

1. 📋 Data Overview

• Visualisasi data pelanggan
• Statistik pendapatan dan churn
• Distribusi segmen pelanggan

2. 📉 Churn Analysis

• Analisis tingkat churn per segmen
• Faktor-faktor penyebab churn
• Korelasi antar variabel

3. 🔮 Predictive Modeling

• Random Forest Classifier
• XGBoost Classifier
• Handling class imbalance
• ROC curves dan feature importance

4. 👥 Customer Segmentation

• K-Means Clustering
• Analisis optimal cluster
• Profil segmen pelanggan

5. 🤖 Explainable AI

• LIME explanations
• SHAP values
• Model interpretability

6. 📈 Strategic Recommendations

• Rekomendasi bisnis
• ROI analysis
• Implementation roadmap

7. 🔧 Model Management

• Save/load trained models
• Model information

🚀 Quick Start

Prerequisites

• Python 3.8+
• pip package manager

Installation & Running

1. Install dependencies:

   pip install -r requirements_iconnet.txt

2. Run the dashboard:

   streamlit run iconnet_dashboard/main.py.py

3. Or use the automated script:

   chmod +x iconnet_dashboard/run_dashboard.sh
   ./iconnet_dashboard/run_dashboard.sh

4. Access the dashboard: Open your browser and go to http://localhost:8501

🔧 Technical Details

Libraries Used

• Streamlit: Web dashboard framework
• Pandas & NumPy: Data manipulation
• Scikit-learn: Machine learning algorithms
• XGBoost: Gradient boosting
• Plotly: Interactive visualizations
• LIME & SHAP: Model interpretability
• Matplotlib & Seaborn: Statistical plots

Machine Learning Models

1. Random Forest: Ensemble method with balanced classes
2. XGBoost: Gradient boosting with scale_pos_weight
3. K-Means: Customer segmentation clustering

Data Features

• Customer demographics and behavior
• Service usage patterns
• Payment and contract information
• Service quality metrics
• Customer satisfaction scores

🎨 UI Features

Responsive Design

• Mobile-friendly layout
• Interactive visualizations
• Professional styling

Navigation

• Sidebar navigation
• Tab-based sections
• Expandable information panels

Error Handling

• Missing file fallbacks
• Graceful error recovery
• User-friendly error messages

📊 Business Impact

Key Metrics

• Churn Rate: Monitor customer retention
• Revenue Analysis: Track monthly revenue patterns
• Segmentation: Identify high-value customers
• Satisfaction: Monitor service quality

Strategic Benefits

• 15% churn reduction potential
• 20% revenue growth opportunities
• Improved customer experience
• Data-driven decision making

🔍 Usage Guide

For Business Users

1. Start with Data Overview for general insights
2. Check Churn Analysis for retention metrics
3. Review Strategic Recommendations for action items

For Technical Users

1. Explore Predictive Modeling for model performance
2. Use Explainable AI for model interpretability
3. Utilize Model Management for saving/loading models

For Data Scientists

1. Analyze Customer Segmentation for clustering insights
2. Examine feature importance and SHAP values
3. Customize models and parameters as needed

🚨 Troubleshooting

Common Issues

1. Logo files missing: Dashboard will use fallback designs
2. LIME/SHAP errors: Check library versions and compatibility
3. Memory issues: Reduce sample sizes for large datasets

Performance Tips

• Use data sampling for large datasets
• Clear browser cache if visualizations don't load
• Restart the dashboard if memory usage is high

👥 Team

• Hani Setiawan (hani.setiawan@binus.ac.id)
• Jetbar Runggu Hamonangan Doloksaribu (jetbar.doloksaribu@binus.ac.id)
• Naufal Yafi (naufal.yafi@binus.ac.id)

📧 Support

Untuk pertanyaan teknis atau dukungan, silakan hubungi tim pengembang.

---

© 2025 PT PLN ICON PLUS SBU Regional Jawa Tengah