Use the normal equation method to fit a regression line, then interpret the result carefully. The main goal here is not only calculation. The main goal is meaning.
The age of the car is the independent variable.
\[ X=\text{Age of car in years} \]
The maintenance cost is the dependent variable because it is the value we want to estimate.
\[ Y=\text{Annual maintenance cost in Rs. '000} \]
The following table gives the age of the cars of a certain company and annual maintenance costs. Obtain the regression equation for cost related to age and estimate the cost of maintenance for a \(10\)-year-old car. Does it match the observed value? Why?
| Age of cars \(X\) in years | 2 | 4 | 6 | 8 | 10 |
|---|---|---|---|---|---|
| Maintenance cost \(Y\) in Rs. '000 | 10 | 15 | 22 | 32 | 46 |
First prepare the table for \(\sum X\), \(\sum Y\), \(\sum X^2\), and \(\sum XY\). Then form the two normal equations and solve for \(A\) and \(B\).
Let age of car be \(X\) and annual maintenance cost be \(Y\).
| \(X\) | \(Y\) | \(X^2\) | \(XY\) |
|---|---|---|---|
| 2 | 10 | 4 | 20 |
| 4 | 15 | 16 | 60 |
| 6 | 22 | 36 | 132 |
| 8 | 32 | 64 | 256 |
| 10 | 46 | 100 | 460 |
| Total | \(\sum X=30\) | \(\sum Y=125\) | \(\sum X^2=220,\quad \sum XY=928\) |
Assume the regression line:
\[ \hat{Y}=A+BX \]
The normal equations are:
\[ \sum Y=nA+B\sum X \]
\[ \sum XY=A\sum X+B\sum X^2 \]
Substitute the values \(n=5,\sum X=30,\sum Y=125,\sum X^2=220,\sum XY=928\).
First normal equation:
\[ 125=5A+30B \]
Second normal equation:
\[ 928=30A+220B \]
Now solve the two equations.
\[ 125=5A+30B \]
Multiplying by \(6\),
\[ 750=30A+180B \]
Also,
\[ 928=30A+220B \]
Subtracting,
\[ 178=40B \]
\[ B=4.45 \]
Substitute \(B=4.45\) in \(125=5A+30B\):
\[ 125=5A+30(4.45) \]
\[ 125=5A+133.5 \]
\[ 5A=-8.5 \]
\[ A=-1.70 \]
Therefore, the regression equation is:
\[ \hat{Y}=-1.70+4.45X \]
This means that for each additional year of car age, the predicted annual maintenance cost increases by about \(4.45\) thousand rupees.
Since the unit of \(Y\) is Rs. '000, \(4.45\) means:
\[ 4.45\times1000=4450 \]
So the predicted maintenance cost increases by about Rs. \(4450\) for each additional year of car age.
For a \(10\)-year-old car, put \(X=10\):
\[ \hat{Y}=-1.70+4.45(10) \]
\[ \hat{Y}=-1.70+44.50 \]
\[ \hat{Y}=42.80 \]
Hence, the estimated annual maintenance cost for a \(10\)-year-old car is:
\[ 42.80\times1000=\text{Rs. }42800 \]
Does it match the observed value?
From the given table, the observed maintenance cost for a \(10\)-year-old car is \(46\) thousand rupees.
\[ 46\times1000=\text{Rs. }46000 \]
The predicted value is Rs. \(42800\), but the observed value is Rs. \(46000\). So it does not match exactly.
This is normal. A regression line gives the best fitted trend using all the data points. It gives an estimated value, not a guaranteed exact value for every observation.
About the negative intercept:
The intercept \(A=-1.70\) is only a mathematical part of the fitted line. It should not be interpreted as a real maintenance cost for a brand-new car. In this problem, the useful interpretation is mainly the slope \(B=4.45\) and the prediction within the given age range.