Search for question
Question

a. What kind of problem does the time plot of Y: depict? How do you suggest correcting for

this problem? Present the time plot of the suggested series.

b. Follow the procedure presented in Unit 6 to model both the seasonality and trend

components of toy sales for the estimation period of 1968.01-1993.12. More specifically,

estimate the following three models and explain which one is the best for forecasting.

Log(Y)= au D1 + 02 D2 + 03 D3 + 04 D4+ as Ds + as Do + a D7 + as Da + as D⁹ +010 D10

where t = 1968.01-1993.12

+ all D11 +012 D12+ U

Log(Y)= a D₁ + a₂D₂ + α3 D3 + 04 D4+ as Ds+a6D6+ a Dr+as Ds+ag Dg + α10 D10

+ α1 D11 +012 D12 + Bi Time + Ur

where t = 1968.01-1993.12

Log(Y) = a1 Di + a2 D2 + 03 D3 + 04 D4+ as Ds + as D6 + a Dr+as Ds + a De+aio Dio

where t = 1968.01-1993.12 (3)

+ ali D11 + a12 D12 + ß: Time + ß2 (Time)² + ur

c. Using the best model, forecast toy sales for the forecast period of 1994.01-1994.12.

Report your forecasts in a table along with the actual values, the forecast errors, the

absolute forecast errors, and the absolute forecast percent errors.

d. Find the mean forecast error (ME), the mean absolute forecast error (MAE), and the mean

absolute percent error (MAPE) and interpret them.

e. Calculate the bias proportion (M) and the variance proportion (US).

f. Plot the forecast error series against the zero-line. Also, plot the actual and forecast series

(together). Using these time plots along with the calculated values of UM and US, comment

on the forecasting performance of the model.

Fig: 1