Whenever you're ready, walk me through how you'd approach this.
How do you identify the root cause of an operational anomaly using high-volume metric data? Explain how you would validate the anomaly, distinguish a real operational change from instrumentation or data-quality issues, decompose the affected KPI, and prioritize hypotheses for investigation. Describe the analyses, visualizations, statistical checks, and operational actions you would use. Assume the data is high volume and may contain seasonality, delayed events, missing records, correlated metrics, and segment-level effects.
Metric time series, event-level logs, dimension attributes, data-pipeline health checks, deployment history, incident records, and dashboard metadata.