R: Error in UseMethod("group_by_") : applied to an object of class

Question

I am working on an assignment where I am supposed to read in a csv file, then send it to a function to be converted to a class object. I managed to read in the csv file and convert it to an object by doing this:

make_LD <- function(x){
  structure(list(id = c(x$id), visit = c(x$visit),
                 room = c(x$room), value = c(x$value), timepoint = c(x$timepoint)), class = "LongitudinalData")
}

A reproducible version of the input CSV file is:

data <- structure(list(id = c(14L, 14L, 14L, 14L, 14L, 14L, 14L, 14L, 
14L, 14L, 14L, 14L, 14L, 14L, 14L, 14L, 14L, 14L, 14L, 14L), 
    visit = c(0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 
    0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L), room = c("bedroom", "bedroom", 
    "bedroom", "bedroom", "bedroom", "bedroom", "bedroom", "bedroom", 
    "bedroom", "bedroom", "bedroom", "bedroom", "bedroom", "bedroom", 
    "bedroom", "bedroom", "bedroom", "bedroom", "bedroom", "bedroom"
    ), value = c(6, 6, 2.75, 2.75, 2.75, 2.75, 6, 6, 2.75, 2.75, 
    2.75, 2.75, 2.75, 2.75, 2.75, 2.75, 2.75, 2.75, 2.75, 2.75
    ), timepoint = 53:72), .Names = c("id", "visit", "room", 
"value", "timepoint"), class = "data.frame", row.names = c(NA, 
-20L))

How I am running this code:

## Read in the data
library(readr)
library(magrittr)
library(dplyr)
source("oop_code_2.R")
## Load any other packages that you may need to execute your code

data <- read_csv("data/MIE.csv")
x <- make_LD(data)
out <- subject(x, 14)

Once I have done this, I use the object variable and send it to a generic function:

subject <- function(x, id) UseMethod("subject")
subject.LongitudinalData <- function(x, subj){
  subj_exist <- x %>%
    group_by_(x$id) %>%
    filter(x$id == subj)
  return(subj_exist)
}

When I run the code, it produces an error:

Error in UseMethod("group_by_") : 
  no applicable method for 'group_by_' applied to an object of class "LongitudinalData"

I have noticed the format of the csv files read in is organized into columns, where as the format of the data after I sent it to become an object has changed into fields.

Question is, what I am doin wrong? Thanks!

Edited/Added:

When I run this code on the data from the csv it works without error, as shown below. If this helps.

> datatest1 <- data %>%
+ group_by(id, visit, room) %>%
+ select(id, visit, room , value) %>%
+ filter(id == 14) %>%
+ summarise(valmean = mean(value))
> print(datatest1)
# A tibble: 6 x 4
# Groups:   id, visit [?]
     id visit         room   valmean
  <int> <int>        <chr>     <dbl>
1    14     0      bedroom  4.786592
2    14     0  living room  2.750000
3    14     1      bedroom  3.401442
4    14     1 family  room  8.426549
5    14     2      bedroom 18.583635
6    14     2  living room 22.550694

When done on the LongitudinalData object, it throws an error:

> datatest2 <- x %>%
+ group_by(id, visit, room) %>%
+ select(id, visit, room , value) %>%
+ filter(id == 14) %>%
+ summarise(valmean = mean(values))
Error in UseMethod("group_by_") : 
  no applicable method for 'group_by_' applied to an object of class "LongitudinalData"

It could also be from the way the data is formatted. Here's an example how the data looks before and after the data is converted to the LongitudinalData object.

> head(data)
# A tibble: 6 x 5
     id visit    room value timepoint
  <int> <int>   <chr> <dbl>     <int>
1    14     0 bedroom  6.00        53
2    14     0 bedroom  6.00        54
3    14     0 bedroom  2.75        55
4    14     0 bedroom  2.75        56
5    14     0 bedroom  2.75        57
6    14     0 bedroom  2.75        58
> head(x)
$id
   [1] 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14
  [40] 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14
  [79] 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14
$visit
   [1] 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
  [60] 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
 [119] 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Here's a link to the data: data

Solution

In setting the class attribute to "LongitudinalData", you're telling R to use only methods for .LongitudinalData. Just like you how you've defined a subject.LongitudinalData that gets called when you execute subject(x, 14), R looks for group_by_.LongitudinalData when you call group_by_, but that, of course, doesn't exist because you just invented the class.

However, R has a simple inheritance-like feature so that you can specify backup classes to try if there is no method for the main class.

From ?class:

When a generic function fun is applied to an object with class attribute c("first", "second"), the system searches for a function called fun.first and, if it finds it, applies it to the object. If no such function is found, a function called fun.second is tried. If no class name produces a suitable function, the function fun.default is used (if it exists). If there is no class attribute, the implicit class is tried, then the default method.

Therefore, you can specify that your LongitudinalData object can also be treated like a data frame like so:

make_LD <- function(x){
  structure(list(id = c(x$id), visit = c(x$visit),
                 room = c(x$room), value = c(x$value), timepoint = c(x$timepoint)), 
  class = c("LongitudinalData", "data.frame"))
}

However, there is some additional structure to a data frame that is missing, so it's usually better to create new classes by building upon an existing object rather than creating one from scratch:

make_LD <- function (x) {
  class(x) <- c("LongitudinalData", class(x))
  x
}

Note that there are several additional issues with your subject.LongitudinalData method that need to be corrected before it will function. I suggest reading vignette("programming", package = "dplyr")

Comments

[MrFlick]

What exactly are you trying to do with group_by_(x$id)? It's not clear what you are trying to do so it's not clear how you are doing it wrong. You probably don't want the underscored version of the function and you probably want to leave out the x$ parts.

[afterafx]

My goal is to group all the entries with the same id and eventually get the mean of their values.

[MrFlick]

It's easier to help if you provide a reproducible example with sample input and desired output.

[afterafx]

Added an example, if that helps.

[Kamil]

Please add a reproducible example where you create the data frame or provide a link to a file that we can download ourselves, @afterafx

[MrFlick]

Sharing a head() is not that helpful. Try a dput(). But it seems like you want your object to act like a data.frame? If that's the case you need to change it's class. use class = c("LongitudinalData", "data.frame") in your make_LD (also show how/when you are calling this function)

[afterafx]

I tried showing the dput() of the data, but it was too large and page became unresponsive so I have included a link to download the data. I also included how I am calling the function make_LD.

[Len Greski]

Just edited the OP adding a dput() with first 20 rows of data. Original data is available in Roger Peng's Biostat 776 GitHub repository.

[afterafx]

It works when using the data directly, but not when the data is converted to the LongitudinalData object. I included the code above that shows how I am using the data.

[PitterJe]

@afterafx which package do you use for this?

[afterafx]

using library(dplyr)

[PitterJe]

@afterafx ok. and in the code for datatest2 from where x comes from? Is it the dataframe you have from the reading process from csv? Because from the first block of code it is not available an x

[afterafx]

If you look at the third block of code, x is generated from: x <- make_LD(data). data is what's read in from the csv file.

[afterafx]

If I change my function to also be treated like a data frame, when I print out class(x), it will say both LongitudinalData and data frame. Where as my assignment calls for it to only state LongitudinalData. Is this still possible?

[W. Murphy]

Then you have two options: define new methods group_by_.LongitudinalDataet cetera to teach each dplyr method how to work with your class, or use base functions instead of dplyr. It's a shame your class is teaching bad habits.

[afterafx]

It seems those are my only options. Your answer worked as far as getting rid of the error.