In: Statistics and Probability
The accompanying table, MultiLinear Regression 5, provides data for tar, nicotine, and carbon monoxide (CO) contents in a certain brand of cigarette. All measurements are in milligrams (mg).
MultiLinear Regression 5
Nicotine (Y) | Tar (X1) | CO (X2) |
0.4 | 5 | 3 |
0.9 | 9 | 11 |
0.7 | 12 | 18 |
0.8 | 13 | 18 |
1 | 16 | 18 |
0.6 | 6 | 6 |
0.9 | 15 | 18 |
1.1 | 15 | 15 |
0.8 | 13 | 18 |
1.2 | 17 | 16 |
Part a) Run the Multilinear Regression Analysis
in Excel with both predictor variables. What is the adjusted
R2 value to the nearest hundredth of a percent (i.e.
45.67%)?
Adjusted R2 =
Part b) What is the p-value for the full model?
Round answer to nearest thousandth of a percent (i.e.
0.123%).
Model p-value =
Part c) What are the p-values for the variable
coefficients? Round answers to the nearest hundredth of a percent
(i.e. 0.12%)
Tar p-value =
CO p-value =
Part d) Run simple linear regression for just
Nicotine & Tar and then for just Nicotine & CO. Write down
the R2
values and p-values for those two models.
Model |
R2 value (hundredth percent, i.e. 12.34%) |
Model p-value (thousandth percent, i.e. 0.123%) |
Nicotine & Tar | ||
Nicotine & CO |
Part e) Using your analyses from parts a)
through d), determine the model that is the best fit for these
data. Use a significance level of 0.05, and consider each model’s
p-value, R2 value (or adjusted R2 value), the
p-values for each of the coefficients to determine the best fit
model. Use that model to estimate the Nicotine level in a cigarette
having 9 mg of tar and 9 mg of CO. Round answer to nearest
hundredth of a milligram (i.e. 0.81 mg).
Predicted Nicotine Level =
Please find Excel Output attached in image
a.)
Adjusted R2 = 76.99%
b.)
p-value for Full Model;
p = 0.00242 = 0.242%
c.) p-value for individual coefficients
For Tar (X1), p = 0.00501 = 0.501%
For CO (X2), p = 0.1431 = 14.31%
d.)
For Nicotine and Tar:
For Nicotine and CO:
Model | Adjusted R-square | P-value |
Nic and tar | 72.05% | 0.11% |
Nic and CO | 33.24% | 4.73% |
e.)
Based on Adjusted R-square and model p-Value, first model looks best, Although variable CO(X2) p-value shows that its insignificant and can be removed
So, For CO(X2) = 9 and Tar (X1) = 9
Predicted Nicotin = 0.71 mg