The AI Issues Finder uses artificial intelligence to automatically detect anomalies, recurring problems, and notable patterns in your forecast data. Instead of manually reviewing thousands of station-days, let the AI surface the issues that matter most.
[IMAGE: ai-issues-finder-overview.png]
Issues View
The default view presents a list of AI-detected issues, each with:
- Severity indicator: Color-coded badges (critical, warning, info) showing issue importance
- Issue title: A brief description of the detected problem
- Category: The type of issue (e.g., systematic bias, outlier pattern, regional degradation)
- Affected stations: Which stations are impacted
- Impact assessment: How this issue affects overall accuracy
- Recommendation: Suggested actions to investigate or resolve the issue
[IMAGE: ai-issue-card.png]
Detailed Data View
Switch to the detailed data view to see the underlying station-level data that the AI analyzed. This table provides the raw accuracy numbers behind each flagged issue, helping you verify and investigate further.
Metric and Observation Day Selectors
Control which data the AI analyzes:
- Metric selector: Focus the analysis on specific forecast variables (temperature, precipitation, etc.)
- Observation day selector: Choose which days to include in the analysis
How the AI Works
The AI Issues Finder runs automated analysis on your forecast data, looking for:
- Systematic biases: Consistent over- or under-forecasting at specific stations or regions
- Sudden degradations: Sharp accuracy drops that may indicate data feed or model problems
- Outlier patterns: Stations with unusually high error rates compared to neighbors
- Recurring issues: Problems that appear repeatedly across multiple analysis periods
Acting on Issues
When you identify a significant issue:
- Review the affected stations in the Daily Insights screen for detailed data
- Add critical stations to your Watchlist for ongoing monitoring
- Use the Data Export tool to pull raw data for internal analysis