转换数据

Question

我有一个关于人们购买特定产品（假设是肥皂）的交易数据，我想了解购买肥皂的强度如何随时间变化。我将强度定义为一天中使用的肥皂的平均数量，假设如果一个人再次购买 he/she 完成 his/her 之前的肥皂供应。也很高兴为每个消费者删除异常值（消费者的平均个人 +- 2 * 消费者的标准差个人）并在上次购买后删除信息

当前数据框如下所示：

transacrions <- data.frame(Client_ID = c(1, 2, 1, 3, 4, 1, 3, 2, 1), 
                           date = c("2017-01-01", "2017-01-01", "2017-01-02", "2017-01-03", "2017-01-04", "2017-01-05", "2017-01-06", "2017-01-09", "2017-01-10"),
                           soaps_bought = c(1, 12, 2, 19, 20, 10, 32, 12, 11))

我认为需要特定步骤才能达到我想要的 table。第一步是 "fill" 每个 he/she 未通过 NA 购买的消费者的缺失日期：

partial_results <- data.frame(Client_ID = rep(1:4, each = 10), 
                              date = rep(seq(as.Date("2017-01-01"), as.Date("2017-01-10"), by = "day"), 4),
                              soaps_bought = c(1,  2,  NA, NA, 10, NA, NA, NA, NA, 11,
                                               12, NA, NA, NA, NA, NA, NA, NA, 12, NA,
                                               NA, NA, 19, NA, NA, 32, NA, NA, NA, NA,
                                               NA, NA, NA, 20, NA, NA, NA, NA, NA, NA ))

第二步将计算两次购买之间经过的天数并计算平均使用量。最好也延迟上次购买：

partial_results_II <- data.frame(Client_ID = rep(1:4, each = 10), 
                              date = rep(seq(as.Date("2017-01-01"), as.Date("2017-01-10"), by = "day"), 4),
                              avg_soaps_bought = c(1/1,  2/3,  2/3,  2/3,  10/5, 10/5, 10/5, 10/5, 10/5, 11/1,
                                                   12/8, 12/8, 12/8, 12/8, 12/8, 12/8, 12/8, 12/8, 12/2, 12/2,
                                                   NA,   NA,   19/3, 19/3, 19/3, 32/5, 32/5, 32/5, 32/5, 32/5,
                                                   NA,   NA,   NA,   20/7, 20/7, 20/7, 20/7, 20/7, 20/7, 20/7 ))

第三步我知道怎么做-它会从长变宽table:

desired_results <- dcast(setDT(partial_results_II), Client_ID ~ date, value.var = "avg_soaps_bought")

第四 - 延迟异常值会延迟消费者 1 的最后日期（使用 11 个肥皂）：我计算了每个人的平均值和标准差，甚至检查了哪些是异常值，但我现在不知道如何根据那个来延迟观察

desired_results_DF <- data.frame(desired_results)
avg <-  rowMeans(desired_results_DF[, -1], na.rm = TRUE)
library(matrixStats)
desired_results_MX <- data.matrix(desired_results_DF[, -1])
sd <-   rowSds(desired_results_MX, na.rm = TRUE)

is_ok <- desired_results_DF[, -1] < avg + 2 * sd |  desired_results_DF[, -1] > avg - 2 * sd

Answer 1

你可以这样做：

library(dplyr)
library(tidyr)
transacrions %>%
  group_by(Client_ID) %>%
  mutate(avg_soaps_bought = soaps_bought/as.numeric(c(diff(as.Date(date)),1))) %>%
  complete(date) %>%
  fill(avg_soaps_bought) %>% # partial_results_II
  select(-soaps_bought) %>%
  spread(date,avg_soaps_bought) # final result

# # A tibble: 4 x 9
# # Groups:   Client_ID [4]
#   Client_ID `2017-01-01` `2017-01-02` `2017-01-03` `2017-01-04` `2017-01-05` `2017-01-06` `2017-01-09` `2017-01-10`
# *     <dbl>        <dbl>        <dbl>        <dbl>        <dbl>        <dbl>        <dbl>        <dbl>        <dbl>
# 1         1          1.0    0.6666667    0.6666667    0.6666667     2.000000          2.0            2           11
# 2         2          1.5    1.5000000    1.5000000    1.5000000     1.500000          1.5           12           12
# 3         3           NA           NA    6.3333333    6.3333333     6.333333         32.0           32           32
# 4         4           NA           NA           NA   20.0000000    20.000000         20.0           20           20

我稍微改变了你的操作顺序，但你第一步想要的功能是 tidyr::complete

转换数据

Transforming data

r

transform

reshape2

dplyr