将数据从宽格式更改为长格式并在 R 中创建计算字段答案

【问题标题】：Change data from wide to long format and create calculated fields in R将数据从宽格式更改为长格式并在 R 中创建计算字段
【发布时间】：2019-04-25 16:12:29
【问题描述】：

以下数据捕获每个 Adv (Adv_Code) 的每月 OPN（最佳产品编号）。 Change_Dt 捕获 Adv 状态从 A 更改为 B 的月份。
变更月份之前，所有 OPN 都属于 adv 的 A 状态，而当月之后，所有 OPN 都属于 B 状态。

以下是现有数据

Adv_Code    Change_Dt   April_OPN   May_OPN June_OPN    July_OPN    Aug_OPN Sep_OPN Oct_OPN Nov_OPN Dec_OPN Jan_OPN Feb_OPN March_OPN
A201        April       0           0       0           0           0       0       0       0       0       0       0       0
A198        July        2           0       0           1           2       0       5       0       0       0       0       0
S1212       Nov         0           3       4           0           0       3       0       1       0       0       0       0

我想通过转换为长格式并根据 OPN 月份创建 Adv_Status 来创建以下数据结构。即如果 Month_OPN 是

Month_OPN 就是从 4 月到 3 月，即 12 个月。
OPN 捕获每个 Adv 的每月 OPN。因此它是每个 Adv 的 April NOP 到 Mar NOP 列中的值的转置。

预期输出：

Agent_Code  Change_Dt   Month_OPN   Adv_Status  OPN                                 
S1198201    April       April       B           0                                   
S1198201    April       May         B           0                                   
S1198201    April       June        B           0                                   
S1198201    April       July        B           0                                   
S1198201    April       Aug         B           0                                   
S1198201    April       Sep         B           0                                   
S1198201    April       Oct         B           0                                   
S1198201    April       Nov         B           0                                   
S1198201    April       Dec         B           0                                   
S1198201    April       Jan         B           0                                   
S1198201    April       Feb         B           0                                   
S1198201    April       Mar         B           0                                   
S1198203    July        April       A           2                                   
S1198203    July        May         A           0                                   
S1198203    July        June        A           0                                   
S1198203    July        July        B           1                                   
S1198203    July        Aug         B           2                                   
S1198203    July        Sep         B           0                                   
S1198203    July        Oct         B           5                                   
S1198203    July        Nov         B           0                                   
S1198203    July        Dec         B           0                                   
S1198203    July        Jan         B           0                                   
S1198203    July        Feb         B           0                                   
S1198203    July        Mar         B           0                                   
S1198212    Nov         April       A           0                                   
S1198212    Nov         May         A           3                                   
S1198212    Nov         June        A           4                                   
S1198212    Nov         July        A           0                                   
S1198212    Nov         Aug         A           0                                   
S1198212    Nov         Sep         A           3                                   
S1198212    Nov         Oct         A           0                                   
S1198212    Nov         Nov         B           1                                   
S1198212    Nov         Dec         B           0                                   
S1198212    Nov         Jan         B           0                                   
S1198212    Nov         Feb         B           0                                   
S1198212    Nov         Mar         B           0

有人可以帮我在 R 中做到这一点吗？

【问题讨论】：

如何在预期中创建“Adv_Status”列
能否解释一下创建该列的逻辑
底层逻辑是捕捉adv状态从A到B的变化时间，可以是Apr到Mar的任意月份，而那个月份的变化由Change_Dt捕捉到。
试试df1 %>% gather(key, OPN, April_OPN:March_OPN) %>% separate(key, into = c("Month_OPN", "key2")) %>% select(-key2) %>% mutate(Adv_Status = case_when(Change_Dt %in% c("April", "Nov") ~ "B", TRUE ~ "A"))
那么与Nov相比的最后几个月不应该是B吗？

标签： r reshape

【解决方案1】：

考虑使用内置常量 month.name 和 month.abb 进行清理和月份数计算的基本 R 的 reshape：

# RESHAPE
rdf <- reshape(df, idvar=c("Adv_Code", "Change_Dt"),
               varying=list(names(df)[-1][-1]), v.names="OPN",
               times=names(df)[-1][-1], timevar="Month_OPN",
               new.row.names=1:1E5, direction="long")

# CALCULATION
final_df <- within(rdf, {    
     # RETRIEVE MONTH NUMBER FROM MONTH NAME/MONTH ABBREV (e.g., JULY or JUL => 7)
     Change_Dt_Num <- sapply(Change_Dt, function(x) max(which(month.name==x), which(month.abb==x)))
     # REMOVE THE "_OPN" SUFFIX FROM Month_OPN VALUES
     Month_OPN <- sub("_OPN", "", Month_OPN)
     # RETRIEVE MONTH NUMBER FROM MONTH NAME/MONTH ABBREV (e.g., JULY or JUL => 7)
     Month_OPN_Num <- sapply(Month_OPN, function(x) max(which(month.name==x), which(month.abb==x)))

     # CONDITIONALLY ASSIGN "A" AND "B" BY COMPARING BOTH MONTH NUMBERS BEFORE/AFTER APRIL
     Adv_Status <- ifelse(Month_OPN_Num < Change_Dt_Num & Month_OPN_Num >= 4, "A", 
                          ifelse(Month_OPN_Num < Change_Dt_Num & Month_OPN_Num < 4, "B", "B"))
     # REMOVE HELPER COLUMNS (USED FOR ABOVE CALCULATION ONLY)    
     rm(Change_Dt_Num, Month_OPN_Num)
})

# RE-ORDER ROWS AND RESET ROW NAMES
final_df <- with(final_df, final_df[order(Adv_Code),])
row.names(final_df) <- NULL

输出

final_df
#    Adv_Code Change_Dt Month_OPN OPN Adv_Status
# 1      A198      July     April   2          A
# 2      A198      July       May   0          A
# 3      A198      July      June   0          A
# 4      A198      July      July   1          B
# 5      A198      July       Aug   2          B
# 6      A198      July       Sep   0          B
# 7      A198      July       Oct   5          B
# 8      A198      July       Nov   0          B
# 9      A198      July       Dec   0          B
# 10     A198      July       Jan   0          B
# 11     A198      July       Feb   0          B
# 12     A198      July     March   0          B
# 13     A201     April     April   0          B
# 14     A201     April       May   0          B
# 15     A201     April      June   0          B
# 16     A201     April      July   0          B
# 17     A201     April       Aug   0          B
# 18     A201     April       Sep   0          B
# 19     A201     April       Oct   0          B
# 20     A201     April       Nov   0          B
# 21     A201     April       Dec   0          B
# 22     A201     April       Jan   0          B
# 23     A201     April       Feb   0          B
# 24     A201     April     March   0          B
# 25    S1212       Nov     April   0          A
# 26    S1212       Nov       May   3          A
# 27    S1212       Nov      June   4          A
# 28    S1212       Nov      July   0          A
# 29    S1212       Nov       Aug   0          A
# 30    S1212       Nov       Sep   3          A
# 31    S1212       Nov       Oct   0          A
# 32    S1212       Nov       Nov   1          B
# 33    S1212       Nov       Dec   0          B
# 34    S1212       Nov       Jan   0          B
# 35    S1212       Nov       Feb   0          B
# 36    S1212       Nov     March   0          B

Rextester Demo

【讨论】：

非常感谢@Parfait。这就像一个魅力！再次感谢您.. :)
你好@Parfait..你能详细解释一下final_df部分吗？我可能需要为此代码 sn-p 创建一个函数，以使用 OPN 以外的变量，因此！
查看使用内嵌 cmets 编辑的代码以解释每个步骤。
非常感谢 cmets @Parfait.. 但是我注意到当 Change_Dt 是 1 月或 2 月或 3 月时逻辑不起作用。每个 FY 都是从 4 月到 3 月，因此逻辑应该适用于所有这些月。
根据您发布的数据，Jan-Mar 值应该是多少？