With time series forecasts, models incorporate various features to improve prediction accuracy. Feature coefficients are critical as you interpret forecast results. 

Explainability coefficients may be generated for:

  • Related time series (RTS)
  • Past values of the target time series (TTS)
  •  Seasonal and trend components

A coefficient with a value greater than zero means there's a positive relationship between explanability data features and the TTS. As the value of the RTS increases, the value of the TTS increases as well. Examples:

  • A positive coefficient for a lagged value suggests that the time series (past values) correlate with future values.
  • A positive coefficient for a seasonal component indicates a positive impact of the season on the TTS.

A coefficient with a value less than zero means a negative relationship between the explainability feature and the TTS. As the value of the RTS increases, the value of the TTS decreases. Examples:

  • A negative coefficient for a competitor's sales means that their increased sales are associated with a decrease in your sales.
  • A negative coefficient for a seasonal component means that the season has a negative impact on the TTS.