Time Series Analysis Forecasting Methods and Evaluation 2 — Questions and Answers
Question 1: What is the 'seasonal naive' forecasting benchmark?
- Using the value from the same period in the prior season as the forecast (Correct answer)
- Using the seasonal mean as the forecast
- Differencing the series and using the naive method
- Using Holt-Winters with default parameters
Correct answer: Using the value from the same period in the prior season as the forecast
Seasonal naive forecasts equal the observation from one full seasonal cycle ago, providing a natural baseline for seasonal data.
Question 2: What does Root Mean Square Error (RMSE) penalize more heavily than MAE?
- Large forecast errors, due to the squaring of residuals (Correct answer)
- Small forecast errors
- Positive errors only
- Negative errors only
Correct answer: Large forecast errors, due to the squaring of residuals
RMSE squares each error before averaging, making it more sensitive to large deviations than MAE, which treats all errors proportionally.
Question 3: What is a 'prediction interval' in forecasting?
- A range within which a future observation will fall with a specified probability (Correct answer)
- The interval between two successive forecasts
- The confidence interval for a model parameter
- The time lag used in forecasting
Correct answer: A range within which a future observation will fall with a specified probability
A prediction interval provides bounds (e.g., 95%) within which a single future observation is expected to fall, accounting for both model uncertainty and noise.
Question 4: What causes a forecast to be 'biased'?
- Systematic over- or under-prediction where the mean error is consistently non-zero (Correct answer)
- High variance in forecast errors
- Large absolute errors
- Non-normality of residuals
Correct answer: Systematic over- or under-prediction where the mean error is consistently non-zero
Bias occurs when a model consistently forecasts too high or too low, resulting in a non-zero mean error (ME) across forecasts.
Question 5: In time series forecasting, what is the 'horizon' h?
- The number of steps ahead being forecast (Correct answer)
- The length of the training data
- The seasonal period of the series
- The number of model parameters
Correct answer: The number of steps ahead being forecast
The forecast horizon h specifies how many time periods into the future the forecast extends (e.g., h=12 for 12-step-ahead monthly forecasts).
Question 6: What is 'forecast reconciliation' in hierarchical time series?
- Adjusting forecasts at different levels so they are mutually consistent (Correct answer)
- Combining multiple model forecasts into one
- Removing bias from individual series forecasts
- Choosing the best model for each series independently
Correct answer: Adjusting forecasts at different levels so they are mutually consistent
Hierarchical reconciliation ensures that forecasts for sub-series sum to forecasts for aggregate series, maintaining logical consistency across levels.
Time Series Analysis Forecasting Methods and Evaluation 2
What is the 'seasonal naive' forecasting benchmark?