Which statement is true about model selection for forecasting?

Which statement is true about model selection for forecasting?

• A. The number of features in the data is irrelevant
• B. The presence of trends, seasonality, and noise level influences model choice
• C. A single universal model fits all datasets
• D. External variables never improve forecasts

Answer: (B)

Forecasting model selection hinges on the data’s characteristics, especially whether there is a trend, seasonality, and how noisy the series is. A trend means the level changes over time, so you use models that can capture or adjust for that movement. Seasonality requires components that repeat at regular intervals. The level of noise tells you how much of the pattern is predictable versus random; high noise suggests simpler approaches to avoid overfitting and to keep forecast intervals realistic, while cleaner signals allow more complex models to extract subtle patterns. External variables can improve forecasts when they carry predictive information, so their presence or absence also shapes the best model choice. This is why the statement that trends, seasonality, and noise influence model choice is true.

Forecasting model selection hinges on the data’s characteristics, especially whether there is a trend, seasonality, and how noisy the series is. A trend means the level changes over time, so you use models that can capture or adjust for that movement. Seasonality requires components that repeat at regular intervals. The level of noise tells you how much of the pattern is predictable versus random; high noise suggests simpler approaches to avoid overfitting and to keep forecast intervals realistic, while cleaner signals allow more complex models to extract subtle patterns. External variables can improve forecasts when they carry predictive information, so their presence or absence also shapes the best model choice. This is why the statement that trends, seasonality, and noise influence model choice is true.