Correlation & Regression • CR11

Question 23: Household Survey Multiple Regression

Fit a multiple regression plane from raw data and interpret each coefficient correctly.

Variables in This Problem

Dependent Variable

\[ Y=\text{Monthly food expenditure in Rs. }100 \]

First Independent Variable

\[ X_1=\text{Monthly income in Rs. }1000 \]

Second Independent Variable

\[ X_2=\text{Size of family} \]

We fit: \[ \hat{Y}=A+BX_1+CX_2 \] by using the three normal equations.
Calculator reminder: because most calculators allow only two-variable statistics mode, collect sums by entering \(X_1,X_2\), then \(Y,X_2\), then \(Y,X_1\). Rename the calculator outputs carefully.

Question 23: Linear Regression Plane

Raw Data • Multiple Regression • Prediction

A household survey on monthly expenditure on food yielded the following data. Find the linear regression plane to predict monthly expenditure. Also predict monthly food expenditure when \(X_1=6\) and \(X_2=8\).

Monthly Expenditure \(Y\) Rs. 10010152025303540
Monthly Income \(X_1\) Rs. 10002457665
Size of family \(X_2\)457108114
✓
Enter your answer first.
Correct answer: \(3.345\)
✓
Enter your answer first.
Correct answer: \(6.796\)
✓
Enter your answer first.
Correct answer: \(-1.761\)
✓
Enter your answer first.
Correct answer: \(30.04\)
✓
Select your answer first.
Correct answer: Predicted expenditure increases by about Rs. \(679.60\) when income increases by Rs. \(1000\), keeping family size constant.
✓
Select your answer first.
Correct answer: The fitted model shows a decrease of about Rs. \(176.10\) for one extra family member, keeping income constant.

Try once before opening the solution

Collect the sums first. Then form the three normal equations and solve for \(A\), \(B\), and \(C\).

Detailed Step-by-Step Solution

The required sums are:

\(n\)7\(\sum Y\)175\(\sum X_1\)35\(\sum X_2\)49
\(\sum X_1^2\)191\(\sum X_2^2\)391\(\sum X_1X_2\)267
\(\sum X_1Y\)945\(\sum X_2Y\)1290

Assume the regression plane:

\[ \hat{Y}=A+BX_1+CX_2 \]

The normal equations are:

\[ \sum Y=nA+B\sum X_1+C\sum X_2 \]

\[ \sum X_1Y=A\sum X_1+B\sum X_1^2+C\sum X_1X_2 \]

\[ \sum X_2Y=A\sum X_2+B\sum X_1X_2+C\sum X_2^2 \]

Substitute the sums:

\[ 175=7A+35B+49C \]

\[ 945=35A+191B+267C \]

\[ 1290=49A+267B+391C \]

Solving these three equations using calculator,

\[ A=3.345,\qquad B=6.796,\qquad C=-1.761 \]

Therefore,

\[ \hat{Y}=3.345+6.796X_1-1.761X_2 \]

For \(X_1=6\) and \(X_2=8\),

\[ \hat{Y}=3.345+6.796(6)-1.761(8)=30.04 \]

Since \(Y\) is measured in Rs. \(100\),

\[ 30.04\times100=\text{Rs. }3004 \]

Interpretation:

\(B=6.796\) means predicted monthly food expenditure increases by about \(6.796\) hundreds of rupees, that is about Rs. \(679.60\), when monthly income increases by Rs. \(1000\), keeping family size constant.

\(C=-1.761\) means this fitted model predicts a decrease of about \(1.761\) hundreds of rupees, that is about Rs. \(176.10\), for one extra family member, keeping income constant. This negative sign should be interpreted cautiously because it comes from the given sample and the relationship between the two independent variables.

Final Answer

Regression plane: \(\hat{Y}=3.345+6.796X_1-1.761X_2\) Predicted expenditure for \(X_1=6,\ X_2=8\) is \(30.04\) hundreds of rupees. Predicted expenditure \(=\text{Rs. }3004\).

Video Solution

In this video, we solve the household survey multiple regression problem by normal equations.
Formulae

Available Formula Sheets

⌂ Home
Latest
Latest Content Auto
Home