Dss Project

DSS Project:

Objective:

Build a firm demand model.

Variables:

For AFD(dependant):

Independents:

Period

Average Demand

Average Price

Average Advertising

Average advertising 1 quarter ago

Average R&D 1 quarter ago

Average advertising 2 quarters ago

Average R&D 2 quarters ago

For NSOM (dependant):

Independents:

Relative Price

Relative Advertising

Normalized Share of Market 1 quarter ago

Relative R&D1 Quarter ago

Relative Advertising 1 Quarter ago

Relative R&D 2 Quarters ago

Relative Advertising 2 Quarters ago

As we will see not all Independent variables are relevant or needed.

Start by Deseasonalizing and trending the deseasonalized values:

Forecasting results for Avg_Dem Date Observation SeasIndex DeseasObs DeseasFCast DeseasError Forecast Error

Q1 1202.000 0.891 1349.574

Moving averages Q2 1482.000 0.878 1688.491

Q3 1814.000 1.150 1577.187

Span 4 Q4 1261.000 1.081 1165.981

Q1 1840.000 0.891 2065.903 1445.308 620.595 1287.266 552.734

Estimation period Q2 1758.000 0.878 2002.947 1624.391 378.557 1425.738 332.262

Deseas Actual Q3 2227.000 1.150 1936.271 1703.005 233.266 1958.709 268.291

MAE 310.0145 299.4436 Q4 2174.000 1.081 2010.185 1792.776 217.409 1938.873 235.127

RMSE 374.0444 354.8475 Q1 1438.000 0.891 1614.548 2003.826 -389.278 1784.711 -346.711

MAPE 14.08% 14.08% Q2 1739.000 0.878 1981.300 1890.988 90.312 1659.732 79.268

Q3 2580.000 1.150 2243.187 1885.576 357.611 2168.693 411.307

Q4 2742.000 1.081 2535.385 1962.305 573.080 2122.218 619.782

Q1 1643.000 0.891 1844.717 2093.605 -248.889 1864.673 -221.673

Q2 1546.000 0.878 1761.409 2151.147 -389.738 1888.076 -342.076

Q3 2306.000 1.150 2004.957 2096.174 -91.217 2410.913 -104.913

Q4 2441.000 1.081 2257.066 2036.617 220.449 2202.586 238.414

Q1 2358.000 0.891 2647.500 1967.037 680.462 1751.945 606.055

Q2 2505.000 0.878 2854.029 2167.733 686.296 1902.633 602.367

Q3 2922.000 1.150 2540.540 2440.888 99.652 2807.385 114.615

Q4 3038.000 1.081 2809.081 2574.784 234.297 2784.609 253.391

Q1 2395.000 0.891 2689.042 2712.787 -23.745 2416.149 -21.149

Q2 2430.000 0.878 2768.579 2723.173 45.406 2390.147 39.853

Trend line for AFD:

Use the trend formula to forecast AFD, then re-seasonalize the forecast by multiplying by the seasonal indices:

AFD= SI*(62.774*PRD+1384.8)

Calculate residuals, then use regression on the residuals

Results of forward regression for Resid

Step 1 - Entering variable: Avg_Price

Summary measures

Multiple R 0.8465

R-Square 0.7165

Adj R-Square 0.7024

StErr of Est 140.9183

ANOVA Table

Source df SS MS F p-value

Explained 1 1003928.8330 1003928.8330 50.5554 0.0000

Unexplained 20 397159.5000 19857.9750

Regression coefficients

Coefficient Std Err t-value p-value Lower limit Upper limit

Constant 15444.5459 2189.7488 7.0531 0.0000 10876.8100 20012.2818

Avg_Price -41.5045 5.8373 -7.1102 0.0000 -53.6809 -29.3282

Step 2 - Entering variable: Avg_Adv

Summary measures Change % Change

Multiple R 0.8786 0.0322 3.8%

R-Square 0.7720 0.0555 7.7%

Adj R-Square 0.7480 0.0456 6.5%

StErr of Est 129.6643 -11.2540 -8.0%

ANOVA Table

Source df SS MS F p-value

Explained 2 1081644.5830 540822.2915 32.1672 0.0000

Unexplained 19 319443.7500 16812.8289

Regression coefficients

Coefficient Std Err t-value p-value Lower limit Upper limit

Constant 13581.4355 2193.3196 6.1922 0.0000 8990.7649 18172.1062

Avg_Price -37.2011 5.7320 -6.4901 0.0000 -49.1982 -25.2040