Cohort Analysis

Introduction

Cohort Analysis 는 고객의 retention/chrun(이탈)에 대한 분석이다. 고객의 트렌드에 대한 이해를 높히며, 소비자 타켓을 구축하거나, 의사 결정에 도움을 주는 분석 기법이다.

UCI 의 Online Retail Data 를 활용하여 분석을 실시한다.

Preprocessing: Features

setwd('C:/Users/Administrator/Desktop/R Analysis/Fast Campus')

Online_Retail <- read_excel("Online Retail.xlsx")
View(Online_Retail)

str(Online_Retail)

## tibble [541,909 x 8] (S3: tbl_df/tbl/data.frame)
##  $ InvoiceNo  : chr [1:541909] "536365" "536365" "536365" "536365" ...
##  $ StockCode  : chr [1:541909] "85123A" "71053" "84406B" "84029G" ...
##  $ Description: chr [1:541909] "WHITE HANGING HEART T-LIGHT HOLDER" "WHITE METAL LANTERN" "CREAM CUPID HEARTS COAT HANGER" "KNITTED UNION FLAG HOT WATER BOTTLE" ...
##  $ Quantity   : num [1:541909] 6 6 8 6 6 2 6 6 6 32 ...
##  $ InvoiceDate: POSIXct[1:541909], format: "2010-12-01 08:26:00" "2010-12-01 08:26:00" ...
##  $ UnitPrice  : num [1:541909] 2.55 3.39 2.75 3.39 3.39 7.65 4.25 1.85 1.85 1.69 ...
##  $ CustomerID : num [1:541909] 17850 17850 17850 17850 17850 ...
##  $ Country    : chr [1:541909] "United Kingdom" "United Kingdom" "United Kingdom" "United Kingdom" ...

Removing Duplicate

중복되는 값들을 제거한다 - unique / duplicate 함수 사용

#--------------------------------------------------------------------
#   총 541909 개의 데이터 중, 중복이 없는 데이터는 536641 개이다. 
#---------------------------------------------------------------------

nrow(unique(Online_Retail)) #536641

## [1] 536641

nrow(Online_Retail) #541909

## [1] 541909

#--------------------------------------------------------------------
#  중복 데이터의 수: 5268개  
#---------------------------------------------------------------------

nrow(Online_Retail)-nrow(unique(Online_Retail))

## [1] 5268

#--------------------------------------------------------------------
#  Duplicated 함수 :  중복값을 제거하고 유일한 값만 선별하기 
#---------------------------------------------------------------------

# 중복된 데이터만 추출 
dupes <- which(duplicated(Online_Retail))


online.retail2 <- Online_Retail[-dupes,]

NA

#--------------------------------------------------------------------
#  Customer ID 결측치 135037개 확인 
#---------------------------------------------------------------------
colSums(is.na(online.retail2))

##   InvoiceNo   StockCode Description    Quantity InvoiceDate   UnitPrice 
##           0           0        1454           0           0           0 
##  CustomerID     Country 
##      135037           0

#--------------------------------------------------------------------
#  결측치를 제거한 데이터 Set 활용 complete.cases 함수 사용 
#---------------------------------------------------------------------


online.retail2[complete.cases(online.retail2),] -> online.retail3

DATES

#--------------------------------------------------------------------
#  as.Date 함수를 사용하여 정리 - Invoice Date 
#---------------------------------------------------------------------

online.retail3$InvoiceDate <- as.Date(online.retail3$InvoiceDate, format="%m/%d/%Y")


#--------------------------------------------------------------------
#  Year 함수를 사용하여 년도 추출 - Invoice Date 
#---------------------------------------------------------------------

online.retail3$year <- as.numeric(format(online.retail3$InvoiceDate, '%Y'))


#--------------------------------------------------------------------
#  2011 년도 데이터만 추출 
#---------------------------------------------------------------------

cohort2011 <- online.retail3[online.retail3$year == 2011,]


#--------------------------------------------------------------------
# Customer ID, Invoice Data, Year 선별을 통해 cohort 그룹 지정 
#---------------------------------------------------------------------


cohort2011 %>% 
  select(CustomerID, InvoiceDate, year) -> cohort2011

Create The Cohorts

1. 각 고객별 InvoiceDate 를 고객들이 JOIN DATE 로 지정

2. Cohort 2011 데이터에 JOIN DATE 와 구매 빈도를 병합 (Inner Join)

3. 고객들의 JOIN DATE의 Month로 각 고객들을 라벨링 한다.

What`s cohort?

cohort 는 특정 시간에 비슷한 성향을 보이는 그룹을 말하는 것으로, 고객들이 실제 고객으로 되는 기준을 날짜로 정해준다. 각 마케팅 채널별로 세분화가 가능할 수 있다. 예를 들어, 온라인데이터에서 제공되는 날짜변수를 통해 월별/주별/분기별/년별 등으로 고객들을 세분화 시키는 분석법을 말한다.

#----------------------------------------------------------------------
#   Cohort2011 데이터에서 각 ID 별 구매 횟수 데이터 프레임 생성 
#--------------------------------------------------------------------


cohort2011 %>% 
  group_by(CustomerID) %>% 
  count() -> order.frequency

#----------------------------------------------------------------------
#   Cohort2011 데이터에서 각 ID 별 최초 구매일 = JOIN DATE/MIN 함수 적용  
#--------------------------------------------------------------------


cohort2011 %>% 
  group_by(CustomerID) %>% 
  summarise(InvoiceDate = min(InvoiceDate)) -> Join.date

#Renaming columns 

colnames(Join.date)[2] <- "Join_Date"


#----------------------------------------------------------------------
#   Cohort2011과 Join_date 병합 
#--------------------------------------------------------------------


merge(cohort2011, Join.date, 
      by.x = "CustomerID",
      by.y = "CustomerID",
      all.x = TRUE) -> cohorts2011

#----------------------------------------------------------------------
#   Month 를 추출하여, Cohort 변수 생성 Ex) JOIN_DATE가 1월이면 라벨링 1 
#--------------------------------------------------------------------


as.numeric(format(cohorts2011$Join_Date, "%m")) -> cohorts2011$Cohort


DT::datatable(head(cohorts2011))

Cohort the days between Invoice and Join Date

가입 날짜 (Join Date) 와 구매 날짜 (Invoice Date) 의 차이를 구해서, Active customer에 대한 cohort 설정

#----------------------------------------------------------------------
#   Invoice - Join = Days difftime 함수 사용 
#--------------------------------------------------------------------


as.numeric(difftime(cohorts2011$InvoiceDate, 
                    cohorts2011$Join_Date,
                    units = c("days"))) -> cohorts2011$days  #Unit 지정: 일 별 계산


#----------------------------------------------------------------------
#  days/30 = month & Invoice Date 일자 삭제 
#--------------------------------------------------------------------


cohorts2011 %>%  
  mutate(month = round(days/30)) %>% 
  mutate(InvoiceDate = format(InvoiceDate, "%Y-%m")) %>% 
  mutate(Join_Date = format(Join_Date, "%Y-%m")) -> df_co




#----------------------------------------------------------------------
#  Cohort 생성하기  
#--------------------------------------------------------------------

c ("Jan Cohort", 
   "Feb Cohort",
   "Mar Cohort",
   "Apr Cohort", 
   "May Cohort", 
   "Jun Cohort",
   "Jul Cohort",
   "Aug Cohort",
   "Sep Cohort",
   "Oct Cohort",
   "Nov Cohort",
   "Dec Cohort") -> groups



for (i in 1:12) {
  
  df_co[df_co$Cohort == i, "Cohort"] <- groups[i]
  
}

#----------------------------------------------------------------------
#   Cohort factor level 고정값 지정  
#--------------------------------------------------------------------
df_co$Cohort <- factor(df_co$Cohort,ordered = T,levels =c("Jan Cohort",
                                                           "Feb Cohort",
                                                            "Mar Cohort",
                                                            "Apr Cohort",
                                                            "May Cohort",
                                                            "Jun Cohort",
                                                            "Jul Cohort",
                                                            "Aug Cohort",
                                                            "Sep Cohort",
                                                             "Oct Cohort",
                                                             "Nov Cohort",
                                                             "Dec Cohort"))
#----------------------------------------------------------------------
#   Cohort factor 변환  
#--------------------------------------------------------------------

#str(cohorts2011$Cohort)

which(duplicated(df_co[-5:-6])) -> dup


df_1 <- df_co[-dup,]


#----------------------------------------------------------------------
# 프레임 변환하기 
#--------------------------------------------------------------------


dcast(df_1, Cohort ~ month, 
      value.var = "CustomerID",
      fun.aggregate = length) ->cohorts.wide



cw.retention <- cohorts.wide
cw.churn <-cohorts.wide




breaks <- quantile(cohorts.wide[,3:13], prob= seq(.05, .95, .05), na.rm = T) 
colors <- sapply(round(seq(155,80, length.out = length(breaks)+1),0),
                 function(x){rgb(x,x,155, maxColorValue = 155)})




datatable(cohorts.wide,
              class = 'cell-border stripe',
             rownames = FALSE,
             options = list(
               ordering=F,
               dom = 't',
               pageLength = 12) ) %>% 
              formatStyle("0",
                         backgroundColor = 'lightgrey',
                         fontWeight = 'bold') %>%
             formatStyle(names(cohorts.wide[c(-1,-2)]),fontWeight = 'bold',color = 'white', 
                         backgroundColor = styleInterval(breaks,colors))

online.retail3 %>%  
  group_by(CustomerID) %>% 
  mutate(Month = min(InvoiceDate),unit="month")

## # A tibble: 401,604 x 11
## # Groups:   CustomerID [4,372]
##    InvoiceNo StockCode Description Quantity InvoiceDate UnitPrice CustomerID
##    <chr>     <chr>     <chr>          <dbl> <date>          <dbl>      <dbl>
##  1 536365    85123A    WHITE HANG~        6 2010-12-01       2.55      17850
##  2 536365    71053     WHITE META~        6 2010-12-01       3.39      17850
##  3 536365    84406B    CREAM CUPI~        8 2010-12-01       2.75      17850
##  4 536365    84029G    KNITTED UN~        6 2010-12-01       3.39      17850
##  5 536365    84029E    RED WOOLLY~        6 2010-12-01       3.39      17850
##  6 536365    22752     SET 7 BABU~        2 2010-12-01       7.65      17850
##  7 536365    21730     GLASS STAR~        6 2010-12-01       4.25      17850
##  8 536366    22633     HAND WARME~        6 2010-12-01       1.85      17850
##  9 536366    22632     HAND WARME~        6 2010-12-01       1.85      17850
## 10 536367    84879     ASSORTED C~       32 2010-12-01       1.69      13047
## # ... with 401,594 more rows, and 4 more variables: Country <chr>, year <dbl>,
## #   Month <date>, unit <chr>

Type of Cohorts

Time Cohorts : 제품과 서비스를 사용하기 시작한 시간 기준 (Month/ Quarter 기준)

Behavior Cohorts : Product 타입 기준 -> Design custome-made Service or product 전략 수립

Size Cohort : 제품과 서비스를 구매한 고객의 사이즈 기준: 특별 기간 동안 쓴 소비액

Online_Retail -> retail


#------------------------------------------------------------------------
#   NA 제거 
#------------------------------------------------------------------------

retail %>%  
  drop_na(CustomerID) -> retail


#------------------------------------------------------------------------
#   중복제거 
#------------------------------------------------------------------------

retail[!duplicated(retail),] ->retail


summary(retail)

##   InvoiceNo          StockCode         Description           Quantity        
##  Length:401604      Length:401604      Length:401604      Min.   :-80995.00  
##  Class :character   Class :character   Class :character   1st Qu.:     2.00  
##  Mode  :character   Mode  :character   Mode  :character   Median :     5.00  
##                                                           Mean   :    12.18  
##                                                           3rd Qu.:    12.00  
##                                                           Max.   : 80995.00  
##   InvoiceDate                    UnitPrice          CustomerID   
##  Min.   :2010-12-01 08:26:00   Min.   :    0.00   Min.   :12346  
##  1st Qu.:2011-04-06 15:02:00   1st Qu.:    1.25   1st Qu.:13939  
##  Median :2011-07-29 15:40:00   Median :    1.95   Median :15145  
##  Mean   :2011-07-10 12:08:23   Mean   :    3.47   Mean   :15281  
##  3rd Qu.:2011-10-20 11:58:30   3rd Qu.:    3.75   3rd Qu.:16784  
##  Max.   :2011-12-09 12:50:00   Max.   :38970.00   Max.   :18287  
##    Country         
##  Length:401604     
##  Class :character  
##  Mode  :character  
##                    
##                    
## 

#------------------------------------------------------------------------
#  0 이상의 데이터만 추출 : MIN 값이 0 인경우 모델링에 negative affect, 제거 
#------------------------------------------------------------------------

retail %>%  
  filter(Quantity >0 & UnitPrice > 0) -> retail

Cohort Analysis Step

1. Invoice Period : year/month 추출

2. Cohort Group : year/month 기반 고객 첫 구매일 추출

3. Cohort period/Cohort Index : 구매일 이후의 month 의 수

library(lubridate)
#------------------------------------------------------------------------
#   ID 별 첫 구매 추출 = retail_cohort
#------------------------------------------------------------------------

retail %>%  
  group_by(CustomerID) %>% 
  summarise(Month = floor_date(min(InvoiceDate), unit = "month")) -> retail_cohort

#------------------------------------------------------------------------
#   InvoiceDate 에서 Month 변수 생성 = retail
#------------------------------------------------------------------------

retail %>% 
  mutate(InvoiceMonth = floor_date(InvoiceDate, unit = "month")) -> retail


#------------------------------------------------------------------------
#   retail_cohort + retail 병합= 각 ID 별 최초 구매 일자를 더해준다. 
#------------------------------------------------------------------------

merge( retail, retail_cohort,
      by.x= "CustomerID",
      by.y = "CustomerID") -> retail_merge



retail_merge %>% 
  mutate(InvoiceMonth_num = month(InvoiceDate),  #구입 한 날의 월/년도
         InvoiceYear_num = year(InvoiceDate),
         CohorMonth_num = month(Month),     #구입 최초 일의 월/년 
         CohortYear_num = year(Month),
         Index = (InvoiceYear_num-CohortYear_num)*12 +(InvoiceMonth_num - CohorMonth_num)+1) -> retail_merge


#------------------------------------------------------------------------
#   최초 구매일, Index 기준 customer의 수 누적 표 
#------------------------------------------------------------------------
retail_merge %>% 
  group_by(Month, Index) %>% #최초구매일, Index
  summarise(Total_Customer = n_distinct(CustomerID)) %>% 
  pivot_wider(names_from = Index, values_from = Total_Customer) %>% 
  rename(DEC = 2, 
         JAN = 3, 
         FEB = 4, 
         MAR = 5, 
         APR = 6,
         MAY = 7, 
         JUN = 8, 
         JUL = 9, 
         AUG = 10, 
         SEP = 11, 
         OCT = 12, 
         NOV = 13,
         DEC_A = 14) -> cohort_counts


#------------------------------------------------------------------------
#   최초 기준 비율 계산 표 
#------------------------------------------------------------------------

cohort_counts %>%  
  mutate(TOTAL = round(DEC/DEC,3)*100,
         JAN_R = round(JAN/DEC,3) *100, 
         FEB_R = round(FEB/DEC,3) *100,
         MAR_R = round(MAR/DEC,3) *100, 
         APR_R = round(APR/DEC,3) *100, 
         MAY_R = round(MAR/DEC,3) *100, 
         JUN_R = round(JUN/DEC,3) *100, 

         JUL_R = round(JUL/DEC,3) *100,
         AUG_R = round(AUG/DEC,3) *100,  
         SEP_R = round(SEP/DEC,3) *100,
         OCT_R = round(OCT/DEC,3) *100,  
         NOV_R = round(NOV/DEC,3) *100,  
         DEC_R = round(DEC_A/DEC,3) *100) %>%  
  select(-c(2:14)) -> retention

retention

## # A tibble: 13 x 14
## # Groups:   Month [13]
##    Month               TOTAL JAN_R FEB_R MAR_R APR_R MAY_R JUN_R JUL_R AUG_R
##    <dttm>              <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
##  1 2010-12-01 00:00:00   100  36.6  32.3  38.4  36.3  38.4  36.3  34.9  35.4
##  2 2011-01-01 00:00:00   100  22.1  26.6  23    32.1  23    24.7  24.2  30  
##  3 2011-02-01 00:00:00   100  18.7  18.7  28.4  27.1  28.4  25.3  27.9  24.7
##  4 2011-03-01 00:00:00   100  15    25.2  19.9  22.3  19.9  26.8  23    27.9
##  5 2011-04-01 00:00:00   100  21.3  20.3  21    19.7  21    21.7  26     7.3
##  6 2011-05-01 00:00:00   100  19    17.3  17.3  20.8  17.3  26.4   9.5  NA  
##  7 2011-06-01 00:00:00   100  17.4  15.7  26.4  23.1  26.4   9.5  NA    NA  
##  8 2011-07-01 00:00:00   100  18.1  20.7  22.3  27.1  22.3  NA    NA    NA  
##  9 2011-08-01 00:00:00   100  20.7  24.9  24.3  12.4  24.3  NA    NA    NA  
## 10 2011-09-01 00:00:00   100  23.4  30.1  11.4  NA    11.4  NA    NA    NA  
## 11 2011-10-01 00:00:00   100  24    11.5  NA    NA    NA    NA    NA    NA  
## 12 2011-11-01 00:00:00   100  11.1  NA    NA    NA    NA    NA    NA    NA  
## 13 2011-12-01 00:00:00   100  NA    NA    NA    NA    NA    NA    NA    NA  
## # ... with 4 more variables: SEP_R <dbl>, OCT_R <dbl>, NOV_R <dbl>, DEC_R <dbl>

#------------------------------------------------------------------------
#  시각화 
#------------------------------------------------------------------------

retention %>% 
  pivot_longer(cols = c(2:14), 
               names_to = "Index", 
               values_to = "Rate") %>% 
  mutate(Index = row_number() %>% as.factor()) %>% 
  arrange(Month) %>% 
  ungroup() %>% 
  mutate(Cohort_Month = ymd(Month) %>%  #ymd 함수를 써야 제대로 나옴 
           as.factor())  -> viz_retention



viz_retention %>% 
  mutate(Cohort_Month = factor(Cohort_Month, levels = rev(levels(Cohort_Month)))) %>% 
  ggplot(aes(Index, Cohort_Month))+
  geom_tile(aes(fill=Rate))+
  geom_text(aes(label=Rate),size = 2)+
  scale_fill_gradientn(
    colours = c("White", "#e6d0a1", "#ff0000"),
    na.value = "White") + 
  labs(title ="Retention Rate (in %)") 

Average Quantity of Each Cohorts

#----------------------------------------------------------------------
#  평균 구매에 대한 Cohort 
#---------------------------------------------------------------------
retail_merge %>% 
  group_by(Month, Index) %>% 
  summarise(avg_quan = round(mean(Quantity),1)) %>% 
  arrange(Month) %>% 
  ungroup() %>% 
  mutate(Cohort_Month = ymd(Month) %>% 
           as.factor(),
         Index = as.factor(Index)) %>% 
  ggplot(aes(x= Index, y=Cohort_Month)) +
  geom_tile(aes(fill=avg_quan))+
  geom_text(aes(label= avg_quan))+
  scale_fill_gradientn(
    colours = c("White", "#188508", "#016000"),
    na.value = c("White"))+
  theme_bw()

RFM with Modelling

R : 최근

F : 자주

M : 금액

#------------------------------------------------------------------------
#   총 금액 구하기 
#------------------------------------------------------------------------

retail_merge %>% 
  mutate(Purchase = (Quantity * UnitPrice)) -> retail_merge


#------------------------------------------------------------------------
#   최초 구매일 
#------------------------------------------------------------------------

max(retail_merge$InvoiceDate)

## [1] "2011-12-09 12:50:00 UTC"

snapshot_date<-max(retail_merge$InvoiceDate) + days(1)
snapshot_date

## [1] "2011-12-10 12:50:00 UTC"

min(retail_merge$InvoiceDate)

## [1] "2010-12-01 08:26:00 UTC"

#------------------------------------------------------------------------
#   RFM
#   R : 마지막 구매 날짜 - 구매 날짜 
#   F : Count
#   M : Quantity * Price 
#------------------------------------------------------------------------


retail_merge %>% 
  group_by(CustomerID) %>% 
  summarise(Recency = as.numeric(round(snapshot_date-max(InvoiceDate),0)),
         Frequency = n(),
         Monetary = sum(Purchase)) -> rfm


head(rfm)

## # A tibble: 6 x 4
##   CustomerID Recency Frequency Monetary
##        <dbl>   <dbl>     <int>    <dbl>
## 1      12346     326         1   77184.
## 2      12347       3       182    4310 
## 3      12348      76        31    1797.
## 4      12349      19        73    1758.
## 5      12350     311        17     334.
## 6      12352      37        85    2506.

#ntile() 함수는 분위수를 계산해주며, n 인자를 통해 몇 분위로 나눌지 선택할 수 있습니다. 해당 함수 역시 #오름차순으로 분위수를 나눕니다

rfm %>% 
  mutate(R = ntile(-Recency,4),
         Fr = ntile(Frequency,4),
         M = ntile(Monetary,4),
         rfm_score= R+Fr+M) -> rfm



#-----------------------------------------------------------------------
#  Gold/Sliver/Bronze 
#---------------------------------------------------------------------------


ifelse(rfm$rfm_score > 9 , "Gold", 
       ifelse(rfm$rfm_score > 5  & rfm$rfm_score <=9, "Sliver", "Bronze")) ->rfm$segment


#-----------------------------------------------------------------------
#  각 segement rfm 평균
#---------------------------------------------------------------------------


rfm %>% 
  group_by(segment) %>% 
  summarise(Recency = mean(Recency),
            Frequency = mean(Frequency),
            Monetary = mean(Monetary)) %>% 
  arrange(desc(Recency)) ->rfm.seg

KMEANS

#-------------------------------------------------
#   Step 1) Scaling 
#--------------------------------------------------


scale(rfm[,2:4]) -> rfm_scale



dist(rfm_scale, method ="euclidean") %>% 
  as.matrix() -> rfm.dist


#-------------------------------------------------
#   Step 2) K 갯수  = 3개로 설정
#--------------------------------------------------


fviz_nbclust(rfm_scale,
             kmeans, 
             method = "wss",
             k.max=9)

fviz_nbclust(rfm_scale,
             kmeans, 
             method = "silhouette",
             k.max=9)

#-------------------------------------------------
#   Step 3) Kmean clustering 
#--------------------------------------------------


kmeans(rfm_scale, 
       centers = 3,
       iter.max = 1000) -> rfm.kmeans

#-------------------------------------------------
#   Step 4) 군집별 평균 값 확인 
#--------------------------------------------------

rfm.kmeans$centers

##      Recency   Frequency    Monetary
## 1 -0.8804868 10.84400785 13.79293005
## 2 -0.5117712  0.04986022 -0.00282326
## 3  1.5463756 -0.28003576 -0.15740003

barplot(t(rfm.kmeans$centers), beside = T, col=2:4)

#-------------------------------------------------
#   Step 5) cluster 할당 
#--------------------------------------------------

as.factor(rfm.kmeans$cluster) -> rfm$cluster_kmean

#-------------------------------------------------
#   Step 6) the characteristics of each cluster
#--------------------------------------------------

rfm %>%  
  group_by(cluster_kmean) %>% 
  summarise(R = mean(Recency),
            F = mean(Frequency),
            M = mean(Monetary)) %>% 
  arrange(desc(R)) -> rfm_kmeans

K-Meoid

set.seed(2021)

#----------------------------------------------------
#   K-medoid 에 부합하는 K 갯수 구하기 
#---------------------------------------------------

fviz_nbclust(rfm_scale, 
             pam, 
             method = "wss",
             k.max=9)

#----------------------------------------------------
#   K-medoid 실행 pam 함수 
#---------------------------------------------------

pam(rfm_scale, k=3) ->rfm.kmedoid


plot(rfm.kmedoid)

barplot(t(rfm.kmedoid$medoids),beside = T, col=2:4)
legend("topright", colnames(rfm_scale), fill=2:14, cex=0.8)

#------------------------------------------------------
#   Clustering 할당 
#-----------------------------------------------------

as.factor(rfm.kmedoid$clustering) -> rfm$medoid_cluster




#------------------------------------------------------
#   Characteristic for each K-medoids clusters
#-----------------------------------------------------

rfm %>% 
  group_by(medoid_cluster) %>% 
  summarise(R = round(mean(Recency),2),
            F = round(mean(Frequency),2),
            M = round(mean(Monetary),2)) %>% 
  arrange(desc(R)) -> rfm_med

DT::datatable(rfm_med)

Cluster 1 : Recency 가 180 이상으로, 6개월 동안 구매 내역이 없는 그룹이다. 반면 빈도는 25, 평균 구매액은 GBP594 Cluster 2 : Recnecy 가 98 으로 3개월 동안 구매 내역이 없는 그룹이다. 빈도수는 41으로, 평균 861 을 구매했다. Cluster 3 : Recency 가 22 으로 마지막 구매일로 부터, 30일 이내 구매ㅎ를 한 그룹으로 142 빈도, 평균 3250 GBP 사용