Create "metadata" field in R - r

Question

I have a data frame set up similar to this:
id <- c(123,234,123,234)
task <- c(54,23,12,58)
a <- c(23,67,45,89)
b <- c(78,45,65,45)
df <- data.frame(id,task,a,b)
> df
id task a b
1 123 54 23 78
2 234 23 67 45
3 123 12 45 65
4 234 58 89 45
where I score a and b for each ID:
df$score <- rowMeans(subset(df, select = c(3:4)), na.rm = TRUE)
> df
id task a b score
1 123 54 23 78 50.5
2 234 23 67 45 56.0
3 123 12 45 65 55.0
4 234 58 89 45 67.0
for each id I got an aggregate score like such:
out <- ddply(df, 1, summarise,
overall = mean(score, na.rm = TRUE))
> out
id overall
1 123 52.75
2 234 61.50
but what I want my final output to have is a new column that has the scores that went into the overall and their task id like this:
id overall meta
1 123 52.75 "task_scores":[{"54":50.5,"12":55}]
2 234 61.50 "task_scores":[{"23":56,"58":67}]
how would I go about doing that using R?

Answer 1

We could make use of jsonlite to create the structure
library(jsonlite)
library(plyr)
ddply(df, "id", summarise, overall = mean(score, na.rm = TRUE),
meta = paste0('"task_scores":',
toJSON(setNames(as.data.frame.list(score), task))))
# id overall meta
#1 123 52.75 "task_scores":[{"54":50.5,"12":55}]
#2 234 61.50 "task_scores":[{"23":56,"58":67}]

Answer 2

I don't know how to make that metadata dictionary offhand, but you could do something like this:
library(dplyr)
library(magrittr)
out <- df %>% group_by(id) %>% mutate(overall = mean(score))
> out
# A tibble: 4 x 6
# Groups: id [2]
id task a b score overall
<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 123 54 23 78 50.5 52.8
2 234 23 67 45 56 61.5
3 123 12 45 65 55 52.8
4 234 58 89 45 67 61.5
So the df would have both the aggregated scores and preserve the data in the original rows.

Answer 3

You can do it with a few mutates. Paste your tallies, get your row average, then your group average.
library(dplyr)
df %>%
mutate(score = rowMeans(subset(., select = c(3:4)), na.rm = TRUE)) %>%
group_by(id) %>%
mutate(overall = mean(score)) %>%
mutate(tally = paste(task, score, sep = ":", collapse = ","))
# A tibble: 4 x 7
# Groups: id [2]
id task a b score overall tally
<dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <chr>
1 123 54 23 78 50.5 52.8 54:50.5,12:55
2 234 23 67 45 56 61.5 23:56,58:67
3 123 12 45 65 55 52.8 54:50.5,12:55
4 234 58 89 45 67 61.5 23:56,58:67
And to get your desired final output, just select and slice.
df %>%
mutate(score = rowMeans(subset(., select = c(3:4)), na.rm = TRUE)) %>%
group_by(id) %>%
mutate(overall = mean(score)) %>%
mutate(tally = paste(task, score, sep = ":", collapse = ",")) %>%
select(id, overall, tally) %>%
slice(1)
# A tibble: 1 x 3
id overall tally
<dbl> <dbl> <chr>
1 123 52.8 54:50.5,12:55
2 234 61.5 23:56,58:67