📞 CallMeMaybe: Operational Efficiency & Telecom Analytics

🎯 Project Objective

The goal of this project was to identify and quantify operational ineffectiveness within the CallMeMaybe virtual telephony service. By integrating call logs and client data, I developed a data-driven framework to classify operator performance based on key metrics like missed calls, waiting times (AWT), and call duration (ACD).

🛠️ Technical Pipeline & Methodology

The project followed a robust data science workflow:

1. Data Integration & Engineering: Unified disparate datasets and engineered the wait_time and is_ineffective features based on business logic.
2. Exploratory Data Analysis (EDA): Identified critical systemic issues, such as a 52.73% missed call rate, far exceeding industry standards.
3. Statistical Validation: Performed Mann-Whitney U tests (non-parametric) to validate if the differences in performance between operator segments were statistically significant.
4. Business Intelligence: Designed and deployed a three-level interactive dashboard in Tableau for real-time monitoring.

📊 Key Findings & Metrics

My analysis revealed significant disparities in operator activity and systemic bottlenecks:

• Average Call Duration (ACD): 26.0 minutes (Industry benchmark: 3-6 min).
• Average Waiting Time (AWT): 100.61 seconds (Ideal: <30 sec).
• Workload Imbalance: A Pareto analysis (80/20) showed that a small percentage of operators handle the majority of the volume, while many remain underutilized.

Statistical Highlights:

• Applied segment classification using percentiles to set fair performance thresholds.
• Validated the "Ineffective" label through hypothesis testing, ensuring recommendations are backed by data, not just intuition.

📈 Interactive Dashboard

The solution features a professional Tableau Dashboard structured for different stakeholders:

• Executive Level: Top-level KPIs for quick decision-making.
• Tactical Level: Pareto charts and individual operator rankings for supervisors.
• Granular Level: Scatter plots and trend lines to identify correlations between wait times and duration.

📁 Enlaces del Proyecto

🌐 Dashboard en Vivo

🔗 View Live Dashboard on Tableau Public

📂 Documentación y Datos

[Carpeta de Google Drive] https://drive.google.com/drive/folders/1vbRrGZHt6WKWrc_Y1ii4uYfKO9xac-wC?usp=sharing

💡 Strategic Recommendations

1. Immediate Deployment: Use the dashboard to redistribute workloads and reduce the current 52% missed call rate.
2. Training Programs: Targeted coaching for operators with high AWT/ACD metrics to align them with industry standards.
3. Incentive Structures: Implement a performance-based bonus system using the efficiency ranking developed in this analysis.

📂 Project Structure

• proyecto_15_telecom.ipynb: Full Python pipeline (Preprocessing + Stats).
• Informe_Ejecutivo_CallMeMaybe.pdf: Comprehensive business report for stakeholders.
• data/: Original and processed datasets (telecom_final_dataset.csv).
• readme.txt: Technical instructions and dashboard links.

🚀 Tech Stack

• Visualization: Tableau Public.
• Data Processing: Python (Pandas, NumPy).
• Statistics: SciPy (Mann-Whitney U Test).
• Documentation: Google Drive / PDF Reporting.

---

Author: Marcel Andrés Palma Céspedes
Role: Data Scientist / Business Intelligence Analyst
Date: 08/18/2025