CTE Data Analytics 2 — Questions and Answers
Question 1: A telecom operator notices its average revenue per user (ARPU) is declining. Which analytics approach best identifies the root cause?
- Segment-level cohort analysis comparing spend patterns over time (Correct answer)
- Increasing marketing spend across all customer tiers
- Reviewing only the top 10% highest-value accounts
- Switching to a new billing platform
Correct answer: Segment-level cohort analysis comparing spend patterns over time
Cohort analysis isolates trends within customer segments, revealing whether ARPU decline stems from churn, plan downgrades, or mix shifts.
Question 2: In telecom network analytics, what does a high packet loss ratio combined with low latency most likely indicate?
- Congestion at a specific node causing drops before queuing pressure builds (Correct answer)
- A billing system misconfiguration
- Insufficient DNS resolution capacity
- High subscriber churn
Correct answer: Congestion at a specific node causing drops before queuing pressure builds
Packet loss with low latency often signals a node-level issue (e.g., hardware fault or misconfigured QoS) rather than broad network saturation.
Question 3: Which statistical technique is most appropriate for forecasting telecom subscriber growth when seasonality is present?
- Simple moving average
- ARIMA with seasonal components (SARIMA) (Correct answer)
- Linear regression without interaction terms
- K-means clustering
Correct answer: ARIMA with seasonal components (SARIMA)
SARIMA explicitly models both trend and seasonal patterns, making it well-suited for subscriber forecasting with cyclical demand.
Question 4: A CTE candidate reviews a dashboard showing NPS scores segmented by geographic region. What additional dimension would most improve actionability?
- Customer service interaction history and resolution rates by region (Correct answer)
- Stock price over the same period
- Total number of employees in each region
- Marketing campaign names
Correct answer: Customer service interaction history and resolution rates by region
Linking NPS to service interaction data reveals whether detractors are driven by specific touchpoint failures that can be operationally addressed.
Question 5: What is 'data normalization' in the context of preparing telecom usage datasets for machine learning?
- Scaling numeric features to a common range so no single feature dominates model training (Correct answer)
- Removing duplicate customer records from the CRM
- Converting all timestamps to UTC
- Encrypting personally identifiable information
Correct answer: Scaling numeric features to a common range so no single feature dominates model training
Normalization rescales features (e.g., call minutes and data usage in different units) so algorithms treat them with equal weight.
Question 6: A carrier's fraud detection system generates many false positives. Which metric should be prioritized to reduce customer impact while maintaining fraud catch rates?
- Precision (Correct answer)
- Recall
- Accuracy
- F-beta score with beta > 1
Correct answer: Precision
Precision measures the share of flagged cases that are truly fraudulent; improving it directly reduces false positives that harm legitimate customers.
Question 7: In telecom capacity planning analytics, what is 'Busy Hour Traffic' (BHT)?
- The peak one-hour period of network load used to size infrastructure capacity (Correct answer)
- Average daily traffic divided by 24
- Maximum theoretical bandwidth of a fiber link
- Traffic during scheduled maintenance windows
Correct answer: The peak one-hour period of network load used to size infrastructure capacity
BHT represents the single busiest hour of traffic and is the standard dimensioning reference to ensure the network handles peak demand.
CTE Data Analytics 2
A telecom operator notices its average revenue per user (ARPU) is declining.
Which analytics approach best identifies the root cause?