SACA Smart Factory Data Analytics 3 — Questions and Answers
Question 1: Which technique is used to detect anomalies in sensor time-series data by modeling the expected data distribution and flagging statistical outliers?
- Moving average smoothing
- Isolation Forest (Correct answer)
- PID control tuning
- Ladder logic parsing
Correct answer: Isolation Forest
Isolation Forest is an unsupervised anomaly detection algorithm well-suited to high-dimensional sensor time-series data in smart factories.
Question 2: In a smart factory context, what does OEE stand for?
- Overall Equipment Effectiveness (Correct answer)
- Operational Efficiency Estimate
- Output Execution Engine
- Optimized Energy Exchange
Correct answer: Overall Equipment Effectiveness
OEE (Overall Equipment Effectiveness) measures how effectively manufacturing equipment is used, combining availability, performance, and quality.
Question 3: A data pipeline ingests 10,000 sensor readings per second. Which storage architecture is best suited for real-time querying and historical trend analysis?
- Relational OLTP database only
- Time-series database with a data lake tier (Correct answer)
- Flat CSV files on local disk
- In-memory cache with no persistence
Correct answer: Time-series database with a data lake tier
A time-series database handles high-frequency ingestion and fast range queries, while a data lake tier stores historical data cost-effectively for trend analysis.
Question 4: What is the primary purpose of a digital twin in smart factory data analytics?
- To replace physical sensors with virtual ones
- To simulate and monitor physical assets using real-time data (Correct answer)
- To encrypt factory network communications
- To manage employee scheduling
Correct answer: To simulate and monitor physical assets using real-time data
A digital twin is a virtual replica of a physical asset that uses real-time sensor data to simulate behavior, enabling monitoring, prediction, and optimization.
Question 5: Which data quality issue occurs when the same event is recorded multiple times in a factory data stream due to network retransmission?
- Schema drift
- Data latency
- Duplicate records (Correct answer)
- Null value propagation
Correct answer: Duplicate records
Duplicate records arise when network retransmission causes the same sensor event to be written more than once, requiring deduplication in the pipeline.
Question 6: In predictive maintenance, a model is trained on vibration data to forecast bearing failure. Which metric best evaluates this model given that missed failures are far more costly than false alarms?
- Accuracy
- Precision
- Recall (Correct answer)
- Specificity
Correct answer: Recall
Recall (sensitivity) measures the proportion of actual failures caught by the model, making it the priority metric when missing a failure is very costly.
Question 7: What does edge computing contribute to smart factory data analytics?
- It centralizes all data processing in a cloud data warehouse
- It processes data close to the source to reduce latency and bandwidth usage (Correct answer)
- It eliminates the need for any cloud infrastructure
- It encrypts data before it reaches the factory floor
Correct answer: It processes data close to the source to reduce latency and bandwidth usage
Edge computing runs analytics near the sensors, enabling millisecond-level response and reducing the volume of raw data sent to the cloud.
SACA Smart Factory Data Analytics 3
Which technique is used to detect anomalies in sensor time-series data by modeling the expected data distribution and flagging statistical outliers?