QME: Lecture5 - 1-4 check

Time series are classified by (matrix classification):

• existence of time trend and seasonality

• existence of time period and seasonality

• existence of random pattern and seasonality

• existence of random pattern and time trend

If Forecast1 has MAD=12; but Forecast2 has MAD=10

• select Forecast 1

• select Forecast 2

• select Forecast 1 and Forecast 2

• select MAD1

If Forecast1 has RMSE=125; but Forecast2 has RMSE=150

• select Forecast 1

• select Forecast 2

• select Forecast 1 and Forecast 2

• none can be selected

If Forecast1 has RMSE=12; but Forecast2 has RMSE=5

• select Forecast 1

• select Forecast 2

• select Forecast 1 and Forecast 2

• none can be selected

If Forecast1 has RMSE=12; but Forecast2 has MAD=5

• select Forecast 1

• select Forecast 2

• select Forecast 1 and Forecast 2

• none can be selected

If the time series looks like this (see the Figure), then consider .. to use for forecasting:

• Exponential smoothing

• Linear time trend

• Classical decomposition

• MAD

If the time series looks like this (see the Figure), then consider .. to use for forecasting:

• Exponential smoothing

• Linear time trend

• Classical decomposition

• MAD

If the time series looks like this (see the Figure), then consider .. to use for forecasting:

• Exponential smoothing

• Linear time trend

• Classical decomposition

• Non-linear time trend

If the time series looks like this (see the Figure), then consider .. to use for forecasting:

• Exponential smoothing

• Linear time trend

• Classical decomposition

• Non-linear time trend

Moving average method for forecasting is ..

• one-period a head

• multi-period a head

• 0-period a head

• 10-period a head