df_customer <- read.csv("df_customer.csv")
head(df_customer)
## X ID_Pelanggan Jenis_Kelamin Tempat_Tinggal Penghasilan Total_Belanja
## 1 1 ID00031 Laki-laki Desa 2227350 2563031
## 2 2 ID00079 Perempuan Kota 9047608 8369550
## 3 3 ID00051 Perempuan Kota 9735540 8053033
## 4 4 ID00014 Laki-laki Kota 13510126 9799876
## 5 5 ID00067 Perempuan Desa 7773498 6982081
## 6 6 ID00042 Laki-laki Desa 6666740 4782002
summary(df_customer)
## X ID_Pelanggan Jenis_Kelamin Tempat_Tinggal
## Min. : 1.00 Length :300 Length :300 Length :300
## 1st Qu.: 75.75 N.unique : 94 N.unique : 2 N.unique : 2
## Median :150.50 N.blank : 0 N.blank : 0 N.blank : 0
## Mean :150.50 Min.nchar: 7 Min.nchar: 9 Min.nchar: 4
## 3rd Qu.:225.25 Max.nchar: 7 Max.nchar: 9 Max.nchar: 4
## Max. :300.00
## Penghasilan Total_Belanja
## Min. : 901314 Min. : 2534171
## 1st Qu.: 6426148 1st Qu.: 5360644
## Median : 8630042 Median : 6552757
## Mean : 8656732 Mean : 6679163
## 3rd Qu.:10921107 3rd Qu.: 7895702
## Max. :16145151 Max. :11626302
nrow(df_customer)
## [1] 300
unique(df_customer$ID_Pelanggan)
## [1] "ID00031" "ID00079" "ID00051" "ID00014" "ID00067" "ID00042" "ID00050"
## [8] "ID00043" "ID00025" "ID00090" "ID00091" "ID00069" "ID00057" "ID00092"
## [15] "ID00009" "ID00093" "ID00099" "ID00072" "ID00026" "ID00007" "ID00083"
## [22] "ID00036" "ID00078" "ID00081" "ID00076" "ID00015" "ID00032" "ID00041"
## [29] "ID00074" "ID00023" "ID00027" "ID00060" "ID00053" "ID00096" "ID00038"
## [36] "ID00089" "ID00034" "ID00063" "ID00013" "ID00082" "ID00097" "ID00021"
## [43] "ID00047" "ID00095" "ID00016" "ID00094" "ID00006" "ID00086" "ID00039"
## [50] "ID00004" "ID00052" "ID00022" "ID00087" "ID00035" "ID00040" "ID00030"
## [57] "ID00012" "ID00064" "ID00071" "ID00085" "ID00037" "ID00008" "ID00098"
## [64] "ID00084" "ID00046" "ID00017" "ID00062" "ID00054" "ID00024" "ID00005"
## [71] "ID00070" "ID00055" "ID00075" "ID00048" "ID00077" "ID00056" "ID00068"
## [78] "ID00001" "ID00088" "ID00020" "ID00049" "ID00059" "ID00011" "ID00066"
## [85] "ID00044" "ID00045" "ID00033" "ID00010" "ID00058" "ID00061" "ID00029"
## [92] "ID00073" "ID00018" "ID00002"
length(unique(df_customer$ID_Pelanggan))
## [1] 94
sort(table(df_customer$ID_Pelanggan), decreasing = TRUE)[1:3]
##
## ID00007 ID00025 ID00089
## 9 7 7
aggregate(Penghasilan ~ Jenis_Kelamin, data = df_customer, mean)
## Jenis_Kelamin Penghasilan
## 1 Laki-laki 8880902
## 2 Perempuan 8505199
aggregate(Penghasilan ~ Tempat_Tinggal, data = df_customer, mean)
## Tempat_Tinggal Penghasilan
## 1 Desa 6249122
## 2 Kota 9878685
aggregate(Total_Belanja ~ Tempat_Tinggal, data = df_customer, mean)
## Tempat_Tinggal Total_Belanja
## 1 Desa 5022231
## 2 Kota 7520118
df_customer[order(-df_customer$Total_Belanja), c("ID_Pelanggan", "Total_Belanja")] |> head(5)
## ID_Pelanggan Total_Belanja
## 76 ID00034 11626302
## 175 ID00011 11527638
## 228 ID00057 11031197
## 287 ID00093 10984825
## 33 ID00007 10846012
table(df_customer$Jenis_Kelamin)
##
## Laki-laki Perempuan
## 121 179
df_customer$Kategori_Penghasilan <- cut(df_customer$Penghasilan,
breaks = c(-Inf, 5000000, 10000000, Inf),
labels = c("Rendah", "Menengah", "Tinggi"))
table(df_customer$Kategori_Penghasilan)
##
## Rendah Menengah Tinggi
## 27 175 98
library(dplyr)
##
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
##
## filter, lag
## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, union
Q1<-df_customer %>%
filter(Total_Belanja>5000000) %>%
count(ID_Pelanggan, sort = TRUE)
Q1
## ID_Pelanggan n
## 1 ID00007 7
## 2 ID00025 7
## 3 ID00026 6
## 4 ID00089 6
## 5 ID00053 5
## 6 ID00079 5
## 7 ID00084 5
## 8 ID00090 5
## 9 ID00093 5
## 10 ID00006 4
## 11 ID00009 4
## 12 ID00016 4
## 13 ID00023 4
## 14 ID00024 4
## 15 ID00031 4
## 16 ID00032 4
## 17 ID00036 4
## 18 ID00057 4
## 19 ID00063 4
## 20 ID00067 4
## 21 ID00074 4
## 22 ID00087 4
## 23 ID00091 4
## 24 ID00002 3
## 25 ID00013 3
## 26 ID00014 3
## 27 ID00030 3
## 28 ID00035 3
## 29 ID00037 3
## 30 ID00039 3
## 31 ID00040 3
## 32 ID00042 3
## 33 ID00045 3
## 34 ID00046 3
## 35 ID00052 3
## 36 ID00055 3
## 37 ID00060 3
## 38 ID00069 3
## 39 ID00071 3
## 40 ID00072 3
## 41 ID00082 3
## 42 ID00085 3
## 43 ID00094 3
## 44 ID00004 2
## 45 ID00008 2
## 46 ID00010 2
## 47 ID00017 2
## 48 ID00021 2
## 49 ID00022 2
## 50 ID00033 2
## 51 ID00034 2
## 52 ID00038 2
## 53 ID00041 2
## 54 ID00048 2
## 55 ID00049 2
## 56 ID00050 2
## 57 ID00051 2
## 58 ID00054 2
## 59 ID00058 2
## 60 ID00066 2
## 61 ID00075 2
## 62 ID00076 2
## 63 ID00083 2
## 64 ID00086 2
## 65 ID00092 2
## 66 ID00096 2
## 67 ID00097 2
## 68 ID00001 1
## 69 ID00005 1
## 70 ID00011 1
## 71 ID00012 1
## 72 ID00015 1
## 73 ID00018 1
## 74 ID00027 1
## 75 ID00029 1
## 76 ID00043 1
## 77 ID00044 1
## 78 ID00047 1
## 79 ID00056 1
## 80 ID00059 1
## 81 ID00061 1
## 82 ID00064 1
## 83 ID00070 1
## 84 ID00073 1
## 85 ID00077 1
## 86 ID00078 1
## 87 ID00088 1
## 88 ID00095 1
## 89 ID00098 1
## 90 ID00099 1
Q2<-df_customer %>%
filter(Jenis_Kelamin == "Perempuan", Tempat_Tinggal == "Kota") %>%
count (ID_Pelanggan, sort = TRUE) %>%
filter(n>5)
Q2
## [1] ID_Pelanggan n
## <0 rows> (or 0-length row.names)
nrow(Q2)
## [1] 0
Q3<-df_customer %>%
filter(Penghasilan > 5000000) %>%
count(ID_Pelanggan, sort = TRUE)
Q3
## ID_Pelanggan n
## 1 ID00007 9
## 2 ID00025 7
## 3 ID00093 7
## 4 ID00026 6
## 5 ID00089 6
## 6 ID00009 5
## 7 ID00014 5
## 8 ID00023 5
## 9 ID00024 5
## 10 ID00053 5
## 11 ID00063 5
## 12 ID00074 5
## 13 ID00079 5
## 14 ID00084 5
## 15 ID00090 5
## 16 ID00006 4
## 17 ID00016 4
## 18 ID00031 4
## 19 ID00032 4
## 20 ID00036 4
## 21 ID00042 4
## 22 ID00046 4
## 23 ID00054 4
## 24 ID00057 4
## 25 ID00067 4
## 26 ID00071 4
## 27 ID00082 4
## 28 ID00085 4
## 29 ID00087 4
## 30 ID00091 4
## 31 ID00002 3
## 32 ID00008 3
## 33 ID00017 3
## 34 ID00030 3
## 35 ID00035 3
## 36 ID00039 3
## 37 ID00040 3
## 38 ID00045 3
## 39 ID00050 3
## 40 ID00052 3
## 41 ID00055 3
## 42 ID00060 3
## 43 ID00069 3
## 44 ID00076 3
## 45 ID00077 3
## 46 ID00083 3
## 47 ID00086 3
## 48 ID00094 3
## 49 ID00097 3
## 50 ID00001 2
## 51 ID00004 2
## 52 ID00005 2
## 53 ID00010 2
## 54 ID00013 2
## 55 ID00020 2
## 56 ID00021 2
## 57 ID00022 2
## 58 ID00027 2
## 59 ID00029 2
## 60 ID00034 2
## 61 ID00037 2
## 62 ID00038 2
## 63 ID00041 2
## 64 ID00043 2
## 65 ID00048 2
## 66 ID00049 2
## 67 ID00051 2
## 68 ID00058 2
## 69 ID00061 2
## 70 ID00064 2
## 71 ID00070 2
## 72 ID00072 2
## 73 ID00075 2
## 74 ID00081 2
## 75 ID00092 2
## 76 ID00096 2
## 77 ID00011 1
## 78 ID00012 1
## 79 ID00015 1
## 80 ID00018 1
## 81 ID00033 1
## 82 ID00044 1
## 83 ID00047 1
## 84 ID00056 1
## 85 ID00059 1
## 86 ID00062 1
## 87 ID00066 1
## 88 ID00068 1
## 89 ID00073 1
## 90 ID00078 1
## 91 ID00088 1
## 92 ID00095 1
## 93 ID00099 1
Q4<-df_customer%>%
filter(Tempat_Tinggal == "Desa", Total_Belanja > 5000000) %>%
count(Jenis_Kelamin, sort = TRUE)
Q4
## Jenis_Kelamin n
## 1 Perempuan 37
## 2 Laki-laki 10
Q5<-df_customer %>%
filter(Tempat_Tinggal == "Desa", Total_Belanja > 5000000) %>%
count(Penghasilan, sort = TRUE)
Q5
## Penghasilan n
## 1 3157783 1
## 2 3373098 1
## 3 3726214 1
## 4 4481204 1
## 5 4626369 1
## 6 4849165 1
## 7 5455465 1
## 8 5575699 1
## 9 5616450 1
## 10 5671820 1
## 11 5809025 1
## 12 5940612 1
## 13 5947963 1
## 14 6005712 1
## 15 6093467 1
## 16 6128487 1
## 17 6193172 1
## 18 6455085 1
## 19 6467267 1
## 20 6499451 1
## 21 6533837 1
## 22 6571179 1
## 23 6776730 1
## 24 6856664 1
## 25 6915574 1
## 26 6928182 1
## 27 6940985 1
## 28 7028460 1
## 29 7082568 1
## 30 7212261 1
## 31 7651846 1
## 32 7773498 1
## 33 7822419 1
## 34 8032910 1
## 35 8108645 1
## 36 8141032 1
## 37 8635642 1
## 38 9024791 1
## 39 9032981 1
## 40 9231091 1
## 41 9310352 1
## 42 9331982 1
## 43 9498495 1
## 44 9657061 1
## 45 9810087 1
## 46 10832415 1
## 47 11664452 1