Spacecloudsky/mtn-telecom-bi-analysis

Python EDA and 5-page Power BI executive dashboard revealing revenue leakage, churn risk, and network quality gaps across 10,000 MTN Nigeria customers. telecom_README.md # MTN Nigeria Telecom Operations Intelligence A two-layer analysis of a 10,000-customer MTN Nigeria telecom dataset, built to help telecom leadership understand customer behaviour, network performance, revenue collection risk, and customer experience — at both the operational and executive level. ## Project Structure This project uses one cleaned dataset, explored at two levels: **Layer 1 — Python EDA (`MTN_Telecom_EDA_Cleaned.ipynb`)** Data cleaning and exploratory analysis in pandas and Plotly, answering 10 granular business questions: customer segmentation, churn risk, top users and revenue contributors, KYC compliance, plan performance, network quality by state, payment behaviour, support demand, loyalty/referrals, and device usage patterns. **Layer 2 — Power BI Dashboard (`Mtn_telecom_powerbi_report.pbix`)** A 5-page executive dashboard built on the same cleaned dataset, focused on operational and strategic questions: revenue trends, payment status across the customer base, network quality and call-drop rates by state, customer satisfaction, support ticket volume, and a deep-dive into the lowest-performing state. The two layers answer different questions at different altitudes — Layer 1 is exploratory and customer-level, Layer 2 is aggregated and decision-ready for leadership. ## Dataset `telecom_customers.csv` — 10,000 simulated MTN Nigeria customer records, including demographics, subscription details, usage metrics (voice, data, SMS), network experience (call drops, quality scores, satisfaction), and payment/behavioural indicators. ## Key Insights - Network quality and call-drop rates vary significantly by state, with several states (Niger, Borno, Kebbi, Edo, Cross River) consistently underperforming. - A third of the customer base sits in "Overdue" or "Unpaid" status, representing a significant revenue collection risk. - No customers in this dataset show zero usage combined with unpaid bills — suggesting low immediate churn risk …