防止dcast聚合数据

Question

I have data like:

rating       title
5            Bean
5            Bean
4            Bean
5            Bean
5            Egg
4            Egg
3            Bacon
2            Bacon

And I want to dcast like

dcast(data, rating ~ title, value.var="rating")

So the titles become the column headers, and the ratings for each title are listed below. However, every time it aggregates them instead, but I don't want this.

Answer 1

read.table(text="rating       title
5            Bean
5            Bean
4            Bean
5            Bean
5            Egg
4            Egg
3            Bacon
2            Bacon", header=TRUE, stringsAsFactors=FALSE) %>%
  dplyr::mutate(id = 1:n()) %>% 
  tidyr::spread(title, rating, fill = 0) %>% 
  dplyr::select(-id)
##   Bacon Bean Egg
## 1     0    5   0
## 2     0    5   0
## 3     0    4   0
## 4     0    5   0
## 5     0    0   5
## 6     0    0   4
## 7     3    0   0
## 8     2    0   0

Answer 2

It can be done with dplyr & tidyverse package :

library(dplyr)
library(tidyverse)

data<-data.frame(rating=c(5,5,4,5,5,4,3,2),
                 title=c("Bean","Bean","Bean","Bean","Egg","Egg","Bacon","Bacon"))

Code :

data%>%mutate(dummy = 1:nrow(data)) %>% 
  spread(title, rating, fill = 0) %>% 
  select(-dummy)%>%t()

OUTPUT :

     [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8]
Bacon    0    0    0    0    0    0    3    2
Bean     5    5    4    5    0    0    0    0
Egg      0    0    0    0    5    4    0    0

Answer 3

I'll provide an alternative using data.table , in the chance that your use of dcast is meaningful. In slight contrast to the other answers, I wonder if you are intending this for presentation vice actual summary, since there is no apparent contextual correlation between the different ratings.

library(data.table)
DT <- fread('rating       title
5            Bean
5            Bean
4            Bean
5            Bean
5            Egg
4            Egg
3            Bacon
2            Bacon')

First we need to assign some "id" that is preserved in the pivot. Similarly, since this is for presentation (and we likely want blanks in the unused spaces vice 0 or NA ), I'll convert the

DT$rating <- as.character(DT$rating)
DT[, id := seq_len(.N), by="title"]
DT
#    rating title id
# 1:      5  Bean  1
# 2:      5  Bean  2
# 3:      4  Bean  3
# 4:      5  Bean  4
# 5:      5   Egg  1
# 6:      4   Egg  2
# 7:      3 Bacon  1
# 8:      2 Bacon  2


dcast(DT, id ~ title, value.var = "rating", fill = "")[,id := NULL,][]
#    Bacon Bean Egg
# 1:     3    5   5
# 2:     2    5   4
# 3:          4    
# 4:          5    

Note that this is not intended for calculations and analysis, merely for presentation. If you want to keep everything numbers, then you'll end up with

# starting with fresh `DT`, no as.character done
DT[, id := seq_len(.N), by="title"]
dcast(DT, id ~ title, value.var = "rating")[,id := NULL,][]
#    Bacon Bean Egg
# 1:     3    5   5
# 2:     2    5   4
# 3:    NA    4  NA
# 4:    NA    5  NA

or optionally use dcast(..., fill=0) to replace the NA s with 0 s.

(In this case, it is still not abundantly clear how the three values on any individual row relate to each other, but perhaps there's meaning in your real data/analysis.)