Eigenvalues
|
Eigenvalues of the Correlation Matrix | ||||
|
|
Eigenvalue |
Difference |
Proportion |
Cumulative |
|
1 |
7.39955623 |
4.56261717 |
0.4353 |
0.4353 |
|
2 |
2.83693906 |
1.08161046 |
0.1669 |
0.6021 |
|
3 |
1.75532860 |
0.68218554 |
0.1033 |
0.7054 |
|
4 |
1.07314306 |
0.12899686 |
0.0631 |
0.7685 |
|
5 |
0.94414620 |
0.19400994 |
0.0555 |
0.8241 |
|
6 |
0.75013626 |
0.25703894 |
0.0441 |
0.8682 |
|
7 |
0.49309732 |
0.02101402 |
0.0290 |
0.8972 |
|
8 |
0.47208330 |
0.04784352 |
0.0278 |
0.9250 |
|
9 |
0.42423977 |
0.11504363 |
0.0250 |
0.9499 |
|
10 |
0.30919614 |
0.15578419 |
0.0182 |
0.9681 |
|
11 |
0.15341195 |
0.02481407 |
0.0090 |
0.9771 |
|
12 |
0.12859787 |
0.03195290 |
0.0076 |
0.9847 |
|
13 |
0.09664498 |
0.02484227 |
0.0057 |
0.9904 |
|
14 |
0.07180271 |
0.01335131 |
0.0042 |
0.9946 |
|
15 |
0.05845140 |
0.02599332 |
0.0034 |
0.9980 |
|
16 |
0.03245808 |
0.03169101 |
0.0019 |
1.0000 |
|
17 |
0.00076708 |
|
0.0000 |
1.0000 |
Scree Plot
According to the eigenvalues, it appears four or five
principal components should be used for the analysis. The first four
eigenvalues were all greater than one. The fifth eigenvalue was close to one, but it
does not explain more of the variability than any one individual variable would
(1/17 � 0.0588).
76.85% of the variability is explained by the first four principal
components, and 82.41% of the variability is explained by the first five
principal components.
The scree plots also suggest we should use the first four principal
components.
The plot starts to slightly level off at the fifth principal
component.
Principal Components
|
|
Prin1 |
Prin2 |
Prin3 |
Prin4 |
|
Sports |
0.034248 |
0.442066 |
-.089001 |
-.271723 |
|
SUV |
0.129805 |
-.224212 |
-.493642 |
-.152413 |
|
Wagon |
-.031771 |
-.019550 |
-.028867 |
0.862708 |
|
Minivan |
0.053535 |
-.207014 |
0.276028 |
-.341532 |
|
AWD |
0.092776 |
-.143740 |
-.551073 |
0.080692 |
|
RWD |
0.117336 |
0.374878 |
0.243227 |
0.075547 |
|
SRP |
0.258893 |
0.345468 |
-.016672 |
0.046504 |
|
DealerCost |
0.257356 |
0.346102 |
-.014498 |
0.051591 |
|
Engine |
0.339550 |
0.002649 |
0.048765 |
-.001305 |
|
Cylinders |
0.326164 |
0.079670 |
0.065943 |
0.056018 |
|
Hp |
0.311395 |
0.234928 |
-.004498 |
0.018717 |
|
Cmpg |
-.306206 |
0.017018 |
0.141743 |
0.004267 |
|
Hwympg |
-.306161 |
0.043666 |
0.247651 |
0.026109 |
|
Weight |
0.331723 |
-.182569 |
-.085671 |
0.014675 |
|
WheelBase |
0.254688 |
-.306301 |
0.284592 |
0.057582 |
|
Length |
0.240991 |
-.269975 |
0.335648 |
0.099536 |
|
Width |
0.288698 |
-.216703 |
0.137418 |
-.085770 |
When looking at the first four
principal components, we noticed some interesting
patterns.
Principal
Component #1:
This principal component appears to be a contrast between Cmpg, and
Hwympg versus SUV, AWD, RWD, SRP, DealerCost, Engine, Cylinders, Hp, Weight,
WheelBase, Length, and Width. It almost seems to be comparing vehicles with
high and low gas mileages. For instance, vehicles that get lower gas
mileages tend to have larger engines, more horsepower and cylinders, and bigger
weights and sizes.
Therefore, this principal component might be suggesting that vehicles
with higher city and highway gas mileages will have lower principal component 1
scores, whereas vehicles with lower city and highway gas mileages may have
higher principal component 1 scores.
Principal Component #2: This principal
component appears to be a contrast between Sports, RWD, SRP, DealerCost,
Cylinders, and Hp versus SUV, Minivan, AWD, Weight, WheelBase, Length, and
Width. Overall, it seems to be a contrast between
Sports cars and other vehicles. This is because sports cars tend to have
higher SRPs and Dealer costs, more Horsepower, and are RWD, which all cause the
principal component 2 scores to increase.
Principal Component #3: This principal component appears to be a contrast between Sports, SUV, AWD, and Weight versus Minivan, RWD, Cylinders, Cmpg, Hwympg, WheelBase, Length, and Width. This could possibly be interpreted as a contrast between minivans and SUVs.
Principal Component #4: This principal component appears to be a contrast between Sports, SUV, Minivan, and Width versus Wagon, AWD, RWD, and Length. A more simple explanation of the contrast in this case could be that it is a contrast between Wagon and the other vehicle types.
Let's check out some plots to help visualize these
patterns....
Plots
![]() |
|
Red = Sports Cars Blue = Other
Vehicles
|
![]() |
|
Red = Cars Blue = Other
Vehicles
|
![]() |
|
Red = SUVs Blue = Other
Vehicles
|
![]() |
|
Red = Wagons Blue = Other
Vehicles
|
![]() |
|
Red = Minivans Blue = Other
Vehicles
|
![]() |
|
Red = Cmpg >=19
(median) Blue = Cmpg < 19
(median)
|