Factory Data Analytics Flashcards

Which machine learning algorithm is most commonly used for anomaly detection in continuous sensor streams?

Answer: Isolation Forest

Isolation Forest isolates anomalies by randomly partitioning data, making it effective for detecting rare events in sensor streams.

What is the key difference between descriptive and diagnostic analytics in a factory setting?

Answer: Descriptive summarizes what happened; diagnostic explains why it happened

Descriptive analytics answers 'what happened' through summaries and dashboards, while diagnostic analytics drills into root causes.

A factory wants to predict equipment failure 48 hours in advance. Which analytics approach is most suitable?

Answer: Predictive analytics using historical sensor data and ML models

Predictive analytics uses historical patterns and machine learning to forecast future events like equipment failures.

Which data quality dimension measures the extent to which data is available when needed?

Answer: Timeliness

Timeliness refers to whether data is available at the time it is needed for decision-making.

In an IoT-enabled factory, what protocol is most widely used for lightweight machine-to-machine messaging?

Answer: MQTT

MQTT (Message Queuing Telemetry Transport) is a lightweight publish-subscribe protocol designed for constrained IoT devices.

What is the purpose of a Pareto chart in factory quality analytics?

Answer: To identify the vital few defect causes that account for most quality issues

A Pareto chart ranks defect causes by frequency, highlighting the 20% of causes responsible for 80% of problems.

Which metric measures the proportion of time a machine is actually producing versus its scheduled production time?

Answer: Machine Availability Rate

Machine Availability Rate specifically measures uptime as a percentage of scheduled production time, excluding planned and unplanned downtime.