数据操作和数据透视

Question

我有一个交易数据集，其中与客户相关的每笔交易都显示在单独的行中，如下所示：

Customer_ID     Transaction_Date    Amount
Cust_1           20-Dec-2020          100
Cust_1           28-Dec-2020          800
Cust_1           05-Jan-2021          300
Cust_2           10-Jan-2021          200
Cust_2           08-Feb-2021          300
Cust_3           15-Feb-2021          500

我尝试将日期转换为名称为“1st_Trans_Date”、“2nd_Trans_Date”等的不同列，R 给了我一个稀疏矩阵，每个唯一日期分配给一列，从而生成 1000 多列。

我希望通过一些计算在不同的列中重新构造这些数据，如下所示：

Customer_ID    1st_Trans_Date    2nd_Trans_Date    3rd_Trans_Date    Total_Trans    Total_Amt    Avg_Amt
Cust_1         20-Dec-2020       28-Dec-2020       05-Jan-2021         3             1200         400
Cust_2         10-Jan-2021       08-Feb-2021                           2              500         250
Cust_3         15-Feb-2021                                             1              500         500 
 

Answer 1

您可以计算每个客户的交易次数，并将数据重塑为宽格式。

library(dplyr)
library(tidyr)

df %>%
  group_by(Customer_ID) %>%
  mutate(row = row_number(), 
         Total_Trans = n()) %>%
  ungroup %>%
  select(-Amount) %>%
  pivot_wider(names_from = row, values_from = Transaction_Date, 
              names_prefix = 'Trans_Date')

# Customer_ID Total_Trans Trans_Date1 Trans_Date2 Trans_Date3
#  <chr>             <int> <chr>       <chr>       <chr>      
#1 Cust_1                3 20-Dec-2020 28-Dec-2020 05-Jan-2021
#2 Cust_2                2 10-Jan-2021 08-Feb-2021 NA         
#3 Cust_3                1 15-Feb-2021 NA          NA         

数据

df <- structure(list(Customer_ID = c("Cust_1", "Cust_1", "Cust_1", 
"Cust_2", "Cust_2", "Cust_3"), Transaction_Date = c("20-Dec-2020", 
"28-Dec-2020", "05-Jan-2021", "10-Jan-2021", "08-Feb-2021", "15-Feb-2021"
), Amount = c(100L, 800L, 300L, 200L, 300L, 500L)), 
class = "data.frame", row.names = c(NA, -6L))

Answer 2

这是data.table 方法

library(data.table)
# Sample data -----
DT <- fread("Customer_ID     Transaction_Date    Amount
Cust_1           20-Dec-2020          100
Cust_1           28-Dec-2020          800
Cust_1           05-Jan-2021          300
Cust_2           10-Jan-2021          200
Cust_2           08-Feb-2021          300
Cust_3           15-Feb-2021          500")
# Sumarise by customer
ans <- DT[, .(paste0(Transaction_Date, collapse = ";"),
              Total_Trans = .N), 
          by = .(Customer_ID)]
# Split transactioen dates
ans[, paste0("Trans_Date", 1:length(tstrsplit(ans$V1, ";"))) := tstrsplit(V1, ";")][,V1 := NULL]
#    Customer_ID Total_Trans Trans_Date1 Trans_Date2 Trans_Date3
# 1:      Cust_1           3 20-Dec-2020 28-Dec-2020 05-Jan-2021
# 2:      Cust_2           2 10-Jan-2021 08-Feb-2021        <NA>
# 3:      Cust_3           1 15-Feb-2021        <NA>        <NA>

Answer 3

为了完整起见，实际生成 OP 要求的列的解决方案：

data <- read.table( text=
"Customer_ID     Transaction_Date    Amount
Cust_1           20-Dec-2020          100
Cust_1           28-Dec-2020          800
Cust_1           05-Jan-2021          300
Cust_2           10-Jan-2021          200
Cust_2           08-Feb-2021          300
Cust_3           15-Feb-2021          500
", header=TRUE)

library(toOrdinal)

data %<>% group_by(Customer_ID) %>%
    mutate(
        TxNo = paste0( sapply(seq(n()),toOrdinal),"_Trans_Date" ),
        Total_Trans = n(),
        Total_Amt = sum(Amount),
        Avg_Amt = mean(Amount)
    )

data %>% pivot_wider(
             id_cols=c("Customer_ID","Total_Trans","Total_Amt","Avg_Amt"),
             names_from=TxNo,
             values_from=Transaction_Date ) %>% print.data.frame

输出：

  Customer_ID Total_Trans Total_Amt Avg_Amt 1st_Trans_Date 2nd_Trans_Date 3rd_Trans_Date
1      Cust_1           3      1200     400    20-Dec-2020    28-Dec-2020    05-Jan-2021
2      Cust_2           2       500     250    10-Jan-2021    08-Feb-2021           <NA>
3      Cust_3           1       500     500    15-Feb-2021           <NA>           <NA>