You are on page 1of 9

#1

Initial Cluster Centers


Cluster
1 2 3 4
Zscore(Calories) -1.60469 -.46842 -.27080 2.10056
Zscore(Protein) -2.82240 1.64640 .70560 -.94080
Zscore(Fat) -1.10877 -.75344 -.39811 2.26690
Zscore(Calcium) .38492 -.38397 4.13968 -.47368
Zscore(Iron) 2.06563 2.40779 .08110 -.26106

#2

Iteration Historya
Change in Cluster Centers
Iteration 1 2 3 4
1 1.987 2.495 2.086 1.803
2 .728 .389 .000 .454
3 1.202 .155 .000 .000
4 .000 .000 .000 .000
a. Convergence achieved due to no or small change in cluster
centers. The maximum absolute coordinate change for any
center is .000. The current iteration is 4. The minimum
distance between initial centers is 4.701.

#3

Cluster Membership
Case Number Food Cluster Distance
1 braised beef 4 .324
2 smoked ham 4 .308
3 roast pork 4 .076
4 simmered pork 4 .176
5 hamburger 2 1.248
6 roast lamb leg 4 1.139
7 beef steak 4 .447
8 roast lamb 4 .574
shoulder
9 roast beef 4 1.508
10 canned chicken 2 1.079
11 canned tuna 2 1.171
12 beef tongue 2 1.008
13 broiled mackerel 2 1.021
14 canned beef 2 1.211
15 veal cutlet 2 .759
16 fried perch 2 1.290
17 baked bluefish 2 1.114
18 fried haddock 2 1.571
19 beef heart 2 2.938
20 broiled chicken 2 .803
21 canned 2 2.011
crabmeat
22 raw clams 1 .530
23 canned clams 1 .530
24 canned shrimp 2 1.444
25 canned 3 1.070
mackerel
26 canned salmon 3 1.210
27 canned sardines 3 2.086

#4

Final Cluster Centers


Cluster
1 2 3 4
Zscore(Calories) -1.48118 -.42960 -.55076 1.32863
Zscore(Protein) -2.35200 .40320 -.15680 -.05880
Zscore(Fat) -1.10877 -.51232 -.51655 1.36746
Zscore(Calcium) .43618 -.30891 2.35414 -.45125
Zscore(Iron) 2.27093 -.24150 -.48916 .03833
#5

Distances between Final Cluster Centers


Cluster 1 2 3 4
1 3.990 4.163 5.006
2 3.990 2.735 2.634
3 4.163 2.735 3.904
4 5.006 2.634 3.904

#6

ANOVA
Cluster Error
Mean Square df Mean Square df F Sig.
Zscore(Calories) 7.335 3 .174 23 42.212 .000
Zscore(Protein) 4.480 3 .546 23 8.205 .001
Zscore(Fat) 7.298 3 .179 23 40.872 .000
Zscore(Calcium) 6.657 3 .262 23 25.398 .000
Zscore(Iron) 3.953 3 .615 23 6.431 .003
The F tests should be used only for descriptive purposes because the clusters have been chosen to
maximize the differences among cases in different clusters. The observed significance levels are not
corrected for this and thus cannot be interpreted as tests of the hypothesis that the cluster means are
equal.

#7

Multiple Comparisons
(I) (J) 95% Confidence
Cluster Cluster Mean Interval
Dependent Number Number Differen Std. Lower Upper
Variable of Case of Case ce (I-J) Error Sig. Bound Bound
Zscore Tukey 1 2 -1.052 *
.315 .014 -1.924 -.180
(Calories) HSD 3 -.930 .381 .096 -1.983 .123
4 -2.810* .330 .000 -3.722 -1.898
2 1 1.052 *
.315 .014 .180 1.924
3 .121 .265 .968 -.613 .855
4 -1.758 *
.185 .000 -2.269 -1.247
3 1 .930 .381 .096 -.123 1.983
2 -.121 .265 .968 -.855 .613
4 -1.879 *
.282 .000 -2.660 -1.098
4 1 2.810 *
.330 .000 1.898 3.722
2 1.758* .185 .000 1.247 2.269
3 1.879 *
.282 .000 1.098 2.660
Zscore(Pr Tukey 1 2 -2.755 *
.559 .000 -4.301 -1.209
otein) HSD 3 -2.195* .675 .017 -4.062 -.328
4 -2.293 *
.584 .004 -3.910 -.677
2 1 2.755* .559 .000 1.209 4.301
3 .560 .470 .638 -.741 1.861
4 .462 .328 .506 -.444 1.368
3 1 2.195 *
.675 .017 .328 4.062
2 -.560 .470 .638 -1.861 .741
4 -.098 .500 .997 -1.482 1.286
4 1 2.293 *
.584 .004 .677 3.910
2 -.462 .328 .506 -1.368 .444
3 .098 .500 .997 -1.286 1.482
Zscore(F Tukey 1 2 -.596 .319 .269 -1.480 .287
at) HSD 3 -.592 .386 .434 -1.660 .475
4 -2.476* .334 .000 -3.401 -1.552
2 1 .596 .319 .269 -.287 1.480
3 .004 .269 1.000 -.740 .748
4 -1.880 *
.187 .000 -2.398 -1.362
3 1 .592 .386 .434 -.475 1.660
2 -.004 .269 1.000 -.748 .740
4 -1.884 *
.286 .000 -2.676 -1.092
4 1 2.476 *
.334 .000 1.552 3.401
2 1.880 *
.187 .000 1.362 2.398
3 1.884* .286 .000 1.092 2.676
Zscore(C Tukey 1 2 .745 .387 .246 -.326 1.816
alcium) HSD 3 -1.918* .467 .002 -3.211 -.625
4 .887 .405 .155 -.233 2.007
2 1 -.745 .387 .246 -1.816 .326
3 -2.663 *
.326 .000 -3.564 -1.762
4 .142 .227 .922 -.486 .770
3 1 1.918* .467 .002 .625 3.211
2 2.663 *
.326 .000 1.762 3.564
4 2.805 *
.347 .000 1.846 3.765
4 1 -.887 .405 .155 -2.007 .233
2 -.142 .227 .922 -.770 .486
3 -2.805 *
.347 .000 -3.765 -1.846
Zscore(Ir Tukey 1 2 2.512 *
.593 .002 .872 4.153
on) HSD 3 2.760* .716 .004 .779 4.741
4 2.233 *
.620 .008 .517 3.948
2 1 -2.512 *
.593 .002 -4.153 -.872
3 .248 .499 .959 -1.133 1.628
4 -.280 .348 .851 -1.241 .682
3 1 -2.760 *
.716 .004 -4.741 -.779
2 -.248 .499 .959 -1.628 1.133
4 -.527 .531 .754 -1.996 .941
4 1 -2.233 *
.620 .008 -3.948 -.517
2 .280 .348 .851 -.682 1.241
3 .527 .531 .754 -.941 1.996
*. The mean difference is significant at the 0.05 level.

#8
Number of Cases in
each Cluster
Cluster 1 2.000
2 14.000
3 3.000
4 8.000
Valid 27.000
Missing .000

#9

Zscore(Calories)
Subset for alpha = 0.05
Cluster Number of Case N 1 2 3
Tukey HSD a,b
1 2 -1.4811842
3 3 -.5507554
2 14 -.4295996
4 8 1.3286287
Sig. 1.000 .977 1.000
Duncan a,b
1 2 -1.4811842
3 3 -.5507554
2 14 -.4295996
4 8 1.3286287
Sig. 1.000 .689 1.000
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 3.884.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I error levels are
not guaranteed.

Zscore(Protein)
Subset for alpha = 0.05
Cluster Number of Case N 1 2
Tukey HSD a,b
1 2 -2.3520023
3 3 -.1568002
4 8 -.0588001
2 14 .4032004
Sig. 1.000 .719
Duncana,b 1 2 -2.3520023
3 3 -.1568002
4 8 -.0588001
2 14 .4032004
Sig. 1.000 .329
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 3.884.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I
error levels are not guaranteed.

Zscore(Fat)
Subset for alpha = 0.05
Cluster Number of Case N 1 2
Tukey HSD a,b
1 2 -1.1087718
3 3 -.5165495
2 14 -.5123193
4 8 1.3674579
Sig. .229 1.000
Duncan a,b
1 2 -1.1087718
3 3 -.5165495
2 14 -.5123193
4 8 1.3674579
Sig. .074 1.000
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 3.884.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I
error levels are not guaranteed.

Zscore(Calcium)
Subset for alpha = 0.05
Cluster Number of Case N 1 2 3
Tukey HSDa,b 4 8 -.4512501
2 14 -.3089133
1 2 .4361807
3 3 2.3541419
Sig. .102 1.000
Duncana,b 4 8 -.4512501
2 14 -.3089133 -.3089133
1 2 .4361807
3 3 2.3541419
Sig. .702 .054 1.000
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 3.884.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I error levels are
not guaranteed.

Zscore(Iron)
Subset for alpha = 0.05
Cluster Number of Case N 1 2
Tukey HSD a,b
3 3 -.4891619
2 14 -.2415033
4 8 .0383346
1 2 2.2709276
Sig. .785 1.000
Duncan a,b
3 3 -.4891619
2 14 -.2415033
4 8 .0383346
1 2 2.2709276
Sig. .386 1.000
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 3.884.
b. The group sizes are unequal. The harmonic mean of the group sizes is used. Type I
error levels are not guaranteed.

#10

The number of clusters used for this k-means cluster analysis is 4 as this is the
minimum number of clusters that provided significant differences for all of the five
standardized variables (calories, protein, fat, calcium, and iron) using one-way ANOVA
in SPSS. This means that all of the variables are significant predictors of the food items’
cluster membership.
The first cluster can be described as foods with very low calories, protein, and
fat, and very high iron content. The second cluster can be described as foods with
average content for all five variables. Meanwhile, the third cluster is characterized as
high-calcium foods and the food items under the fourth cluster contain high calories and
fat.
Based on the results of multiple comparisons using Tukey HSD, there is a
significant difference in the calories of food items between all clusters, except 1 – 3 and
2 – 3. For the protein content, food items under cluster 1 is significantly different from
other clusters, while other cluster comparison yielded no significant differences. For the
fat content, food items under cluster 4 is significantly different from other clusters, while
other cluster comparison yielded no significant differences. For the calcium content,
food items under cluster 3 is significantly different from other clusters, while other
cluster comparison yielded no significant differences. Lastly, the iron content of food
items under cluster is significantly different from other clusters, while other cluster
comparison yielded no significant differences.

You might also like