Data Loading

Data is about PDA 2001

# Cluster Analysis for segmentation: MKTG 580/MKTG 880
# R code by Bharatendra Rai
# Modifed by Suh
# Example Data from MEXL: Decisionpro.biz

## package you need to install "cluster"
# install.packages("cluster")
# library(cluster)
##


#install.packages("cluster")
library(cluster)
## Warning: package 'cluster' was built under R version 4.2.3
#getwd()  #identify the current working directory once you open RStudio
#setwd("~/MKTG580-880/Module3")  # change the working directory to the folder where the data is saved
#getwd()  #confirm the current working directory


segdata <- read.csv("pda2001_original.csv")
str(segdata)
## 'data.frame':    160 obs. of  33 variables:
##  $ id          : int  1 2 3 4 5 6 7 8 9 10 ...
##  $ innovator   : int  4 5 3 3 4 5 5 3 6 2 ...
##  $ usemessage  : int  1 4 5 1 1 3 5 1 5 2 ...
##  $ usecell     : int  5 7 5 6 6 5 5 5 5 7 ...
##  $ usepim      : int  7 5 6 7 7 7 6 5 7 5 ...
##  $ infpassive  : int  4 6 3 5 7 3 6 6 3 4 ...
##  $ infactive   : int  6 5 4 7 5 7 4 5 7 5 ...
##  $ remoteaccess: int  3 6 7 3 5 4 3 4 1 5 ...
##  $ shareinf    : int  1 5 6 4 5 6 3 3 3 4 ...
##  $ monitor     : int  4 4 4 4 2 5 4 4 2 6 ...
##  $ email       : int  5 7 4 4 7 6 6 7 6 4 ...
##  $ web         : int  5 3 4 7 5 7 5 4 4 6 ...
##  $ mmedia      : int  6 6 6 5 5 6 7 7 6 5 ...
##  $ ergonomic   : int  4 1 3 5 4 3 4 1 3 3 ...
##  $ monthly     : int  15 15 35 15 25 40 20 15 15 25 ...
##  $ price       : int  280 310 370 390 410 250 310 230 240 290 ...
##  $ age         : int  48 43 48 28 54 62 53 48 46 59 ...
##  $ education   : int  3 3 2 3 3 2 3 2 3 2 ...
##  $ income      : int  47 89 24 52 84 46 71 39 87 21 ...
##  $ construction: int  0 0 0 0 0 0 0 0 0 0 ...
##  $ emergency   : int  0 0 0 0 0 0 0 0 0 0 ...
##  $ sales       : int  0 0 0 1 0 1 1 1 0 0 ...
##  $ service     : int  0 1 0 0 0 0 0 0 0 1 ...
##  $ professional: int  0 0 0 0 0 0 0 0 1 0 ...
##  $ computers   : int  0 0 1 0 1 0 0 0 0 0 ...
##  $ pda         : int  1 0 1 1 1 0 0 0 0 1 ...
##  $ cell        : int  1 1 1 1 1 1 0 1 1 1 ...
##  $ pc          : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ away        : int  3 3 6 7 2 4 6 3 4 6 ...
##  $ businessweek: int  0 1 0 0 1 0 0 0 0 0 ...
##  $ pcmagazine  : int  0 0 0 0 0 0 0 0 0 0 ...
##  $ fieldstream : int  0 0 0 0 0 0 0 0 0 0 ...
##  $ mgourmet    : int  0 0 0 0 0 0 0 0 0 0 ...
head(segdata)
##   id innovator usemessage usecell usepim infpassive infactive remoteaccess
## 1  1         4          1       5      7          4         6            3
## 2  2         5          4       7      5          6         5            6
## 3  3         3          5       5      6          3         4            7
## 4  4         3          1       6      7          5         7            3
## 5  5         4          1       6      7          7         5            5
## 6  6         5          3       5      7          3         7            4
##   shareinf monitor email web mmedia ergonomic monthly price age education
## 1        1       4     5   5      6         4      15   280  48         3
## 2        5       4     7   3      6         1      15   310  43         3
## 3        6       4     4   4      6         3      35   370  48         2
## 4        4       4     4   7      5         5      15   390  28         3
## 5        5       2     7   5      5         4      25   410  54         3
## 6        6       5     6   7      6         3      40   250  62         2
##   income construction emergency sales service professional computers pda cell
## 1     47            0         0     0       0            0         0   1    1
## 2     89            0         0     0       1            0         0   0    1
## 3     24            0         0     0       0            0         1   1    1
## 4     52            0         0     1       0            0         0   1    1
## 5     84            0         0     0       0            0         1   1    1
## 6     46            0         0     1       0            0         0   0    1
##   pc away businessweek pcmagazine fieldstream mgourmet
## 1  1    3            0          0           0        0
## 2  1    3            1          0           0        0
## 3  1    6            0          0           0        0
## 4  1    7            0          0           0        0
## 5  1    2            1          0           0        0
## 6  1    4            0          0           0        0
names(segdata)
##  [1] "id"           "innovator"    "usemessage"   "usecell"      "usepim"      
##  [6] "infpassive"   "infactive"    "remoteaccess" "shareinf"     "monitor"     
## [11] "email"        "web"          "mmedia"       "ergonomic"    "monthly"     
## [16] "price"        "age"          "education"    "income"       "construction"
## [21] "emergency"    "sales"        "service"      "professional" "computers"   
## [26] "pda"          "cell"         "pc"           "away"         "businessweek"
## [31] "pcmagazine"   "fieldstream"  "mgourmet"

column (variable) selection for cluster analysis

# column (variable) selection for cluster analysis
segdata1 <- segdata[, 1:16]
names(segdata1)
##  [1] "id"           "innovator"    "usemessage"   "usecell"      "usepim"      
##  [6] "infpassive"   "infactive"    "remoteaccess" "shareinf"     "monitor"     
## [11] "email"        "web"          "mmedia"       "ergonomic"    "monthly"     
## [16] "price"

Normalizing data and getting distance measures

# Normalize 
z = segdata1[, -1]  # remove ID variable which is not numeric
head(z)
##   innovator usemessage usecell usepim infpassive infactive remoteaccess
## 1         4          1       5      7          4         6            3
## 2         5          4       7      5          6         5            6
## 3         3          5       5      6          3         4            7
## 4         3          1       6      7          5         7            3
## 5         4          1       6      7          7         5            5
## 6         5          3       5      7          3         7            4
##   shareinf monitor email web mmedia ergonomic monthly price
## 1        1       4     5   5      6         4      15   280
## 2        5       4     7   3      6         1      15   310
## 3        6       4     4   4      6         3      35   370
## 4        4       4     4   7      5         5      15   390
## 5        5       2     7   5      5         4      25   410
## 6        6       5     6   7      6         3      40   250
means = apply(z,2,mean)
means
##    innovator   usemessage      usecell       usepim   infpassive    infactive 
##      3.47500      4.20625      5.55625      4.00625      4.45000      4.50000 
## remoteaccess     shareinf      monitor        email          web       mmedia 
##      3.98750      3.71250      4.79375      4.72500      4.47500      4.01250 
##    ergonomic      monthly        price 
##      4.63125     28.75000    332.25000
sds = apply(z,2,sd)
sds
##    innovator   usemessage      usecell       usepim   infpassive    infactive 
##     1.605611     1.690137     1.180272     2.035833     1.624498     1.590063 
## remoteaccess     shareinf      monitor        email          web       mmedia 
##     1.859829     1.634485     1.496734     1.818666     1.999843     1.974643 
##    ergonomic      monthly        price 
##     1.592521     9.696657   113.231271
nor = scale(z,center=means,scale=sds)  # mean 0 and sd 1

##calculate distance matrix (default is Euclidean distance)
distance = dist(nor)
#distance2 = dist(segdata1[,-1])
#distance
#distance2

Cluster Analysis

# Hierarchical clustering using default complete linkage 
segdata.hclust = hclust(distance)
plot(segdata.hclust)

plot(segdata.hclust,labels=segdata$ID,main='Default from hclust')

plot(segdata.hclust,hang=-1)

Interpretation of Cluster Analysis

Selection of number of clusters

Based on eyeball checking, I think there are 4 distinct clusters.

Cluster membership

# Cluster membership
member = cutree(segdata.hclust,4)
member
##   [1] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1
##  [38] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 3 3 3 3 3 1 3 3 3 3 3 3 3 3 3 3 1
##  [75] 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2 2 1 2 1 2 2
## [112] 2 2 1 2 1 2 2 2 1 1 2 2 2 2 2 2 1 2 1 2 1 1 1 1 2 1 2 2 2 1 2 2 2 4 4 4 4
## [149] 4 4 4 4 4 4 4 4 4 4 4 4
table(member)
## member
##  1  2  3  4 
## 71 27 46 16
segdata2 <- cbind(segdata, member)  #adding member variable into the original segdata
write.csv(segdata2,'pda2001_with_member.csv')

Brief description about the results.

Cluster 1 has 71 respondents (biggest one)

Profiling and Naming

#Characterizing clusters 
aggregate(nor,list(member),mean)
##   Group.1  innovator usemessage    usecell      usepim infpassive  infactive
## 1       1  0.2217136 -0.3053652  0.1134407  0.73027010  0.1998451  0.3321685
## 2       2  1.0881976 -0.2096860  0.5014929 -0.03945514 -0.3910040 -0.7104305
## 3       3 -0.7020217  0.8040573 -0.1029147 -0.94275599 -0.4375934 -0.4785152
## 4       4 -0.8018753 -0.6027616 -1.0537823 -0.46356953  1.0310878  1.1005850
##   remoteaccess    shareinf    monitor      email        web     mmedia
## 1   -0.2583344 -0.16017020 -0.1727353  0.5848970  0.5371901  0.5357539
## 2   -0.9690758 -0.41325739 -0.5798119  0.5381436  0.6884801  0.7064100
## 3    0.7197367 -0.02360802  0.5590070 -0.8648239 -0.7049467 -0.8760528
## 4    0.7124313  1.47600018  0.1378000 -1.0172289 -1.5188694 -1.0508230
##    ergonomic    monthly      price
## 1 -0.2195009 -0.2560052 -0.2611818
## 2  0.9525135  0.3580845  0.7880463
## 3 -0.4100348 -0.4091504 -0.5382405
## 4  0.5455186  1.7080629  1.3766073
aggregate(segdata1[,-c(1,1)],list(member),mean)
##   Group.1 innovator usemessage  usecell   usepim infpassive infactive
## 1       1  3.830986   3.690141 5.690141 5.492958   4.774648  5.028169
## 2       2  5.222222   3.851852 6.148148 3.925926   3.814815  3.370370
## 3       3  2.347826   5.565217 5.434783 2.086957   3.739130  3.739130
## 4       4  2.187500   3.187500 4.312500 3.062500   6.125000  6.250000
##   remoteaccess shareinf  monitor    email      web   mmedia ergonomic  monthly
## 1     3.507042 3.450704 4.535211 5.788732 5.549296 5.070423  4.281690 26.26761
## 2     2.185185 3.037037 3.925926 5.703704 5.851852 5.407407  6.148148 32.22222
## 3     5.326087 3.673913 5.630435 3.152174 3.065217 2.282609  3.978261 24.78261
## 4     5.312500 6.125000 5.000000 2.875000 1.437500 1.937500  5.500000 45.31250
##      price
## 1 302.6761
## 2 421.4815
## 3 271.3043
## 4 488.1250
mean2 = apply(segdata1, 2, mean)
mean2
##           id    innovator   usemessage      usecell       usepim   infpassive 
##     80.50000      3.47500      4.20625      5.55625      4.00625      4.45000 
##    infactive remoteaccess     shareinf      monitor        email          web 
##      4.50000      3.98750      3.71250      4.79375      4.72500      4.47500 
##       mmedia    ergonomic      monthly        price 
##      4.01250      4.63125     28.75000    332.25000
aggregate(segdata2[,-c(1,1)],list(member),mean)
##   Group.1 innovator usemessage  usecell   usepim infpassive infactive
## 1       1  3.830986   3.690141 5.690141 5.492958   4.774648  5.028169
## 2       2  5.222222   3.851852 6.148148 3.925926   3.814815  3.370370
## 3       3  2.347826   5.565217 5.434783 2.086957   3.739130  3.739130
## 4       4  2.187500   3.187500 4.312500 3.062500   6.125000  6.250000
##   remoteaccess shareinf  monitor    email      web   mmedia ergonomic  monthly
## 1     3.507042 3.450704 4.535211 5.788732 5.549296 5.070423  4.281690 26.26761
## 2     2.185185 3.037037 3.925926 5.703704 5.851852 5.407407  6.148148 32.22222
## 3     5.326087 3.673913 5.630435 3.152174 3.065217 2.282609  3.978261 24.78261
## 4     5.312500 6.125000 5.000000 2.875000 1.437500 1.937500  5.500000 45.31250
##      price      age education   income construction  emergency     sales
## 1 302.6761 42.16901  2.591549 68.53521   0.05633803 0.01408451 0.4647887
## 2 421.4815 39.03704  3.222222 85.59259   0.03703704 0.03703704 0.1481481
## 3 271.3043 36.47826  2.152174 58.41304   0.06521739 0.02173913 0.2173913
## 4 488.1250 42.18750  1.937500 52.43750   0.31250000 0.18750000 0.0625000
##     service professional computers       pda      cell     pc     away
## 1 0.1267606    0.1126761 0.2112676 0.5070423 0.9014085 1.0000 4.112676
## 2 0.0000000    0.3703704 0.4074074 0.8888889 0.8888889 1.0000 2.481481
## 3 0.3913043    0.1739130 0.1304348 0.1521739 0.9130435 1.0000 4.956522
## 4 0.1250000    0.0000000 0.3125000 0.1875000 0.6250000 0.8125 5.375000
##   businessweek pcmagazine fieldstream   mgourmet member
## 1    0.3098592  0.1971831  0.05633803 0.01408451      1
## 2    0.5185185  0.2962963  0.00000000 0.07407407      2
## 3    0.1739130  0.2826087  0.23913043 0.00000000      3
## 4    0.0000000  0.2500000  0.31250000 0.00000000      4

Screen Plot

# Scree Plot
wss <- (nrow(nor)-1)*sum(apply(nor,2,var))
for (i in 2:20) wss[i] <- sum(kmeans(nor, centers=i)$withinss)
plot(1:20, wss, type="b", xlab="Number of Clusters", ylab="Within groups sum of squares") 

Analysis (comment)

Based on screen plot, 4 (or 3, or 5) cluster solution seems reasonable

Assignment Question 1

I would say, there are 32 distinct segmentations throughout the market #### Assignment Question 2 Within the cluster diagram, there are 4 big clusters that connect them all. #### Assignment Question 3 Through different codes, it gives all the information you need. #### Assignment Question 4 There is a lot of information to take in, so condensing it would be helpful.