Multiple Linear Regression in Industry
Translated by
ScribdTranslationsMultiple Linear Regression in Industry
Translated by
ScribdTranslationsIndustrial Statistics
Mr. Francisco Antonio Torres Espriú
Examples – Multiple Linear Regression
I. In a company dedicated to anodizing aluminum articles (bakitchen utensils), anodizing is achieved
with a solution made from acids (sulfuric, citric, boric) and aluminum dichromate. In this
the pH of the solution, the temperature, the current, and the residence time are controlled.
Due to the thinness of the anodizing, complaints about the lack of resistance and durability have increased.
of the product. To solve this problem, it is decided to study, through an experiment, the relationship of the
pH and temperature with the thickness of the anodizing. The data is shown in the following table:
pH temperature Thickness
1.2 -8 9
1.8 -8 14
1.2 8 10
1.8 8 19
1.2 -8 8
1.8 -8 12
1.2 8 11
1.8 8 20
1.5 0 14
1.5 0 13
a) What are the independent variables and which is the dependent one? Argue.
b) Adjust a delta model Y = β 0+ Beta 1 X1+ beta
2 X2+ε and I wrote down the equation of the fitted model.
c) From the fitted model, what is the estimated thickness when a pH = 2 is used and a
Temperature of 10 degrees? Is the quality of the fit satisfactory? Argue.
d) Is the model appropriate? Argue based on residual graphs, hypothesis testing, and
coefficients of determination.
e) Do you think it would be worth considering adding another term to the model to improve the fit?
Arguments
Industrial Statistics
Mr. Francisco Antonio Torres Espriú
II. Consider the daysall that appear in the following table:
y x1 x2
6 10 3
9 12 11
8 12 4
3 4 1
10 12 11
4 6 1
5 8 7
2 2 4
11 18 8
9 9 10
10 17 8
2 2 5
a) Adjust the following modelY = β 0+ beta1 X1+ β 2X+2 ε that is, find the estimators of
least squares for these regression coefficients.
b) From the adjusted model, estimate the mean response when x1 = 8 and x2 = 7; is this value
different from what was observed under the same conditions? If so, why does this happen?
c) Make the interval estimate for the mean response in the previous point.
d) Build a prediction interval for a future observation with x1 = 8 and x2 = 7.
e) Explain the differences between the two previous intervals.
f) Are the previous estimates adequate? Argue based on the goodness of fit.
model.