引導式測驗 · 中等 等級

時間序列分解測驗

測試附加成分、每月季節性週期、STL 和移動平均值、穩健擬合、未來洩漏以及分解為何具有描述性。

相關引導路徑時間序列分解
問題 1 的 8

In the guide's additive view, how is an observed time-series value represented?

預覽每個問題8 問題

當您參加測驗時,就會出現答案和解釋。

  1. In the guide's additive view, how is an observed time-series value represented?
    • Trend multiplied by the number of observations
    • Trend plus seasonal component plus remainder
    • Seasonal component minus the entire history
    • A future forecast with no residual component
  2. For monthly observations with a repeating yearly cycle, what period should an analyst consider?
    • Seven observations
    • One observation
    • Thirty observations
    • Twelve observations
  3. Which component is intended to capture a slower movement in the series level?
    • The remainder
    • The seasonal position
    • The trend
    • The forecast error from a future model
  4. A December peak repeats after accounting for a rising baseline. Which component should represent that repetition?
    • The training-set size
    • The seasonal component at the chosen yearly period
    • A permanent structural break by definition
    • Only the long-run trend
  5. How does STL differ from the classical moving-average decomposition named in the guide?
    • STL guarantees a causal explanation of seasonality
    • Classical decomposition never estimates a trend
    • STL requires every seasonal peak to be identical forever
    • STL uses local LOESS smoothing and offers robust fitting
  6. Why can a centered trend smoother cause leakage in a forecasting test?
    • It makes the seasonal period equal to zero
    • It necessarily deletes the last observation
    • It may use observations later than the forecast origin, including test-period values
    • It prevents the model from seeing any training history
  7. What does robust STL fitting do with exceptional residual outliers?
    • It can reduce their influence on estimated trend and seasonality
    • It guarantees no future shocks will occur
    • It sets all unusual observations exactly to zero
    • It proves the outliers were data-entry errors
  8. What conclusion does a decomposition plot alone fail to establish?
    • A causal reason for the pattern or a guaranteed future repeat
    • That a method produced estimated components
    • That a series was observed at particular times
    • That a selected seasonal period was used