r使用dplyr的“收集”功能

Question

I have a dataframe that looks like the picture showned below in 'input'.

I try to get 1 date per row (see picture below in 'desired output'). In other word, I try to do a kind of 'transpose' for each row.

Let's stipulate that the combination 'LC' and 'Prod' is a unique key.

Input

Desired output:

Info:

In my real dataset, there is some missing values in the quantity field (the colored region area). Thus, I should still be able to compute with missing values.

My try that fails

I have tried the following but it fails...

library("dplyr")
outputTest <- tbl_df(inputTest) %>%
  gather(date, value, c(inputTest$LC, inputTest$Prod))

outputTest

Source:

inputTest <- structure(list(LC = structure(c(1L, 3L, 1L, 2L), .Label = c("berlin", 
                                                            "munchen", "stutgart"), class = "factor"), Prod = structure(c(1L, 
                                                                                                                          2L, 2L, 1L), .Label = c("(STORE1)400096", "STORE2_00154"), class = "factor"), 
               PROD_TYPE = structure(c(1L, 2L, 2L, 1L), .Label = c("STORE1", 
                                                                   "STORE2"), class = "factor"), X2015.6.29 = c(20.08, 8.91, 
                                                                                                                11.38, 15.42), X2015.7.6 = c(20.66, 8.49, 10.91, 15.57), 
               X2015.7.13 = c(19.02, 8.55, 10.89, 14.6), X2015.7.20 = c(18.6, 
                                                                        7.95, 10.58, 14.31)), .Names = c("LC", "Prod", "PROD_TYPE", 
                                                                                                         "2015.6.29", "2015.7.6", "2015.7.13", "2015.7.20"), class = "data.frame", row.names = c(NA, 
                                                                                                                                                                                                     -4L))

Answer 1

Using gather, you can specify the columns you do not want to gather with the negation operator '-' (minus sign). The key in your case is the date, the value is the value, and LC, Prod, and PROD_TYPE serve as identifiers.

output <- as.data.frame(inputTest) %>%
        tidyr::gather(key = Date, value = Value, -LC, -Prod, -PROD_TYPE)

This yields:

         LC           Prod PROD_TYPE      Date Value
1    berlin (STORE1)400096    STORE1 2015.6.29 20.08
2  stutgart   STORE2_00154    STORE2 2015.6.29  8.91
3    berlin   STORE2_00154    STORE2 2015.6.29 11.38
4   munchen (STORE1)400096    STORE1 2015.6.29 15.42
5    berlin (STORE1)400096    STORE1  2015.7.6 20.66
6  stutgart   STORE2_00154    STORE2  2015.7.6  8.49
7    berlin   STORE2_00154    STORE2  2015.7.6 10.91
8   munchen (STORE1)400096    STORE1  2015.7.6 15.57
9    berlin (STORE1)400096    STORE1 2015.7.13 19.02
10 stutgart   STORE2_00154    STORE2 2015.7.13  8.55
11   berlin   STORE2_00154    STORE2 2015.7.13 10.89
12  munchen (STORE1)400096    STORE1 2015.7.13 14.60
13   berlin (STORE1)400096    STORE1 2015.7.20 18.60
14 stutgart   STORE2_00154    STORE2 2015.7.20  7.95
15   berlin   STORE2_00154    STORE2 2015.7.20 10.58
16  munchen (STORE1)400096    STORE1 2015.7.20 14.31

Answer 2

It is better to have column names that starts as non-numeric. According to ?gather , the ... specifies for selection of columns by using its name. Here, we are interested in the columns that starts with number ie the date columns, so we can use matches and specify a regex to select those columns

library(dplyr)
library(tidyr)
inputTest %>%
       tbl_df %>% 
       gather(date, value, matches("^\\d+") )