The Long Term Trends screen reveals how your forecast accuracy evolves over multiple months. Use this view to identify improvement trajectories, seasonal patterns, and long-term performance shifts.
[IMAGE: long-term-trends-overview.png]
Trend Charts
The main visualization shows your accuracy metrics plotted over time as a line chart. A reference line shows the market average, making it easy to see whether you’re performing above or below the competitive baseline.
Stat Selector
Choose which statistical measure to track over time:
- Avg Error: Mean signed error (shows bias direction)
- Abs Error: Mean absolute error (overall accuracy)
- RMS Error: Root mean square error (penalizes large errors)
- No Error: Percentage of perfect forecasts
- Within ±3°F: Percentage of forecasts within 3 degrees of observed
- Over 10°F: Percentage of forecasts with errors over 10 degrees
[IMAGE: stat-selector.png]
Variance vs. Market Average
A key feature of this screen is the variance display, which shows your deviation from the market average over time. Positive values indicate you’re outperforming the market; negative values indicate underperformance.
Reading the Charts
Look for these patterns in your trend data:
- Consistent improvement: A downward trend in error metrics (or upward in accuracy percentages) suggests your forecast model improvements are working
- Seasonal cycles: Many providers see accuracy fluctuations tied to seasons. Summer convection and winter storms often increase errors
- Step changes: Sudden improvements or degradations may correlate with model updates or data source changes
- Market convergence: Your line approaching the market average may indicate competitors are improving
Filters
Use the standard filter bar to control the trend analysis:
- Days Out: See how trends differ at various lead times
- Location: Focus on specific geographic regions
- Market: Change your competitive comparison group
- Metric: Switch between temperature, precipitation, wind, and sky metrics
Tips
- Compare trends across different days-out ranges to see if improvements are consistent across lead times
- Use this screen to prepare performance review presentations by selecting the most relevant stat and time period