dplyr: manipulate function with multiple arguments by groups - r

With the robCompositions package, I need to impute missing values on a group basis. For example, with the iris dataset.
library(robCompositions)
library(dplyr)
data(iris)
# Insert random NAs
for (i in 1:4) {
n_NA = sample(0:10, 1)
index_NA = sample(1:nrow(iris), n_NA)
iris[index_NA, i] = NA
}
This is where I have no idea which manip to use...
impfunc <- function(x) x %.%
regroup(list(...)) %.%
mutate(impKNNa(x[,-5], k=6, metric="Euclidean"))
impfunc(iris, "Species")
iris %.% group_by(Species) %.% mutate(impKNNa(iris[,-5], k=6, metric="Euclidean"))
Any idea?
Thanks.

Use the the do() function. It allows you to apply any arbitrary function to a grouped data frame.
You'll also want to extract not just the output from impKNNa but specifically impKNNA$xImp which is the altered data frame.
The other issue is that impKNNA doesn't want any variables except the numeric variables of interest and do() won't remove the categorical variables. So perhaps a solution is to write a wrapper function for impKNNA that will remove categorical variables and return xIMP, and use do() to apply that to a grouped data frame.

Related

Apply R glm predict function on dataframe by group

At the moment I am trying to apply GLM predict on a dataframe. The dataframe is quite large therefore I want to apply predict by chunks.
I have found a solution but it is quite unhandy. I first create an empty dataframe and then use rbind. Is there a more efficient way of doing this?
df=data[c(),]
for (x in split(data, factor(sort(rank(row.names(data))%%10)))) {
x["prediction"]=predict(model, x, type="response")
df=rbind(df,x)
}
As the comments mention, an example of what you want your output dataframe to look like would be very helpful.
But I think you can achieve what you want by making a grouping variable first then using 'group_by', something like this:
df <- data %>%
mutate(group = rep(1:10, times = nrow(.)/10)) %>% # make an arbitrary grouping factor for this example
group_by(group) %>% # group by whatever your grouping factor is
summarise(predictions = predict(model, x, type = 'response')) # summarise could be replaced by mutate

Issue computing AUC with pROC package

I'm trying to use a function that calls on the pROC package in R to calculate the area under the curve for a number of different outcomes.
# Function used to compute area under the curve
proc_auc <- function(outcome_var, predictor_var) {
pROC::auc(outcome_var, predictor_var)}
To do this, I am intending to refer to outcome names in a vector (much like below).
# Create a vector of outcome names
outcome <- c('outcome_1', 'outcome_2')
However, I am having problems defining variables to input into this function. When I do this, I generate the error: "Error in roc.default(response, predictor, auc = TRUE, ...): 'response' must have two levels". However, I can't work out why, as I reckon I only have two levels...
I would be so happy if anyone could help me!
Here is a reproducible code from the iris dataset in R.
library(pROC)
library(datasets)
library(dplyr)
# Use iris dataset to generate binary variables needed for function
df <- iris %>% dplyr::mutate(outcome_1 = as.numeric(ntile(Sepal.Length, 4)==4),
outcome_2 = as.numeric(ntile(Petal.Length, 4)==4))%>%
dplyr::rename(predictor_1 = Petal.Width)
# Inspect binary outcome variables
df %>% group_by(outcome_1) %>% summarise(n = n()) %>% mutate(Freq = n/sum(n))
df %>% group_by(outcome_2) %>% summarise(n = n()) %>% mutate(Freq = n/sum(n))
# Function used to compute area under the curve
proc_auc <- function(outcome_var, predictor_var) {
pROC::auc(outcome_var, predictor_var)}
# Create a vector of outcome names
outcome <- c('outcome_1', 'outcome_2')
# Define variables to go into function
outcome_var <- df %>% dplyr::select(outcome[[1]])
predictor_var <- df %>% dplyr::select(predictor_1)
# Use function - first line works but not last line!
proc_auc(df$outcome_1, df$predictor_1)
proc_auc(outcome_var, predictor_var)
outcome_var and predictor_var are dataframes with one column which means they cannot be used directly as an argument in the auc function.
Just specify the column names and it will work.
proc_auc(outcome_var$outcome_1, predictor_var$predictor_1)
You'll have to familiarize yourself with dplyr's non-standard evaluation, which makes it pretty hard to program with. In particular, you need to realize that passing a variable name is an indirection, and that there is a special syntax for it.
If you want to stay with the pipes / non-standard evaluation, you can use the roc_ function which follows a previous naming convention for functions taking variable names as input instead of the actual column names.
proc_auc2 <- function(data, outcome_var, predictor_var) {
pROC::auc(pROC::roc_(data, outcome_var, predictor_var))
}
At this point you can pass the actual column names to this new function:
proc_auc2(df, outcome[[1]], "predictor_1")
# or equivalently:
df %>% proc_auc2(outcome[[1]], "predictor_1")
That being said, for most use cases you probably want to follow #druskacik's answer and use standard R evaluation.

Lagged values multiple columns with function in R

I would like to create lagged values for multiple columns in R.
First, I used a function to create lead/lag like this:
mleadlag <- function(x, n, ts_id) {
pos <- match(as.numeric(ts_id) + n, as.numeric(ts_id))
x[pos]
}
Second, I would like to apply this function for several columns in R. firm.characteristics is list of columns I would like to compute lagged values.
library(dplyr)
firm.characteristics <- colnames(df)[4:6]
for(i in 1:length(firm.characteristics)){
df <- df %>%
group_by(company) %>%
mutate(!!paste0("lag_", i) := mleadlag(df[[i]] ,-1, fye)) %>%
ungroup()
}
However, I didn't get the correct values. The output for all companies in year t is the last row in year t-1. It didn't group by the company any compute the lagged values.
Can anyone help me which is wrong in the loop? Or what should I do to get the correct lagged values?
Thank you so much for your help.
Reproducible sample could be like this:
set.seed(42) ## for sake of reproducibility
n <- 6
dat <- data.frame(company=1:n,
fye=2009,
x=rnorm(n),
y=rnorm(n),
z=rnorm(n),
k=rnorm(n),
m=rnorm(n))
dat2 <- data.frame(company=1:n,
fye=2010,
x=rnorm(n),
y=rnorm(n),
z=rnorm(n),
k=rnorm(n),
m=rnorm(n))
dat3 <- data.frame(company=1:n,
fye=2011,
x=rnorm(n),
y=rnorm(n),
z=rnorm(n),
k=rnorm(n),
m=rnorm(n))
df <- rbind(dat,dat2,dat3)
I would try to stay away from loops in the tidyverse. Many of the tidyverse applications that would traditionally require loops already exist and are very fast, which creates more efficient and intuitive code (the latter being my opinion). This is a great use case for dplyr's across() functionality. I first changed the df to a tibble.
df %>%
as_tibble() %>%
group_by(company) %>%
mutate(
across(firm.characteristics, ~lag(., 1L))
) %>%
ungroup()
This generates the required lagged values. For more information see dplyr's across documentation.

How can you obtain the group_by value for use in passing to a function?

I am trying to use dplyr to apply a function to a data frame that is grouped using the group_by function. I am applying a function to each row of the grouped data using do(). I would like to obtain the value of the group_by variable so that I might use it in a function call.
So, effectively, I have-
tmp <-
my_data %>%
group_by(my_grouping_variable) %>%
do(my_function_call(data.frame(x = .$X, y = .$Y),
GROUP_BY_VARIABLE)
I'm sure that I could call unique and get it...
do(my_function_call(data.frame(x = .$X, y = .$Y),
unique(.$my_grouping_variable))
But, it seems clunky and would inefficiently call unique for every grouping value.
Is there a way to get the value of the group_by variable in dplyr?
I'm going to prematurely say sorry if this is a crazy easy thing to answer. I promise that I've exhaustively searched for an answer.
First, if necessary, check if it's a grouped data frame: inherits(data, "grouped_df").
If you want the subsets of data frames, you could nest the groups:
mtcars %>% group_by(cyl) %>% nest()
Usually, you won't nest within the pipe-chain, but check in your function:
your_function(.x) <- function(x) {
if(inherits(x, "grouped_df")) x <- nest(x)
}
Your function should then iterate over the list-column data with all grouped subsets. If you use a function within mutate, e.g.
mtcars %>% group_by(cyl) %>% mutate(abc = your_function_call(.x))
then note that your function directly receives the values for each group, passed as class structure. It's a bit difficult to explain, just try it out and debug your_function_call step by step...
You can use groups(), however a SE version of this does not exist so I'm unsure of its use in programming.
library(dplyr)
df <- mtcars %>% group_by(cyl, mpg)
groups(df)
[[1]]
cyl
[[2]]
mpg

Use plyr to apply functions stored in lists

I'd like to use plyr to calculate multiple empirical cumulative distribution functions using ecdf(), and then apply those functions appropriately to entries in a data frame. For instance:
# Use the diamonds dataset in ggplot2
library(diamonds)
library(plyr)
# Calculate an ecdf for each combination of cut and color
all_ecdfs <- dlply(diamonds, c("cut", "color"), function(x) ecdf(x$carat))
# Make a dataset of specific diamonds, which I want to compare to the larger set
# My particular subset of diamonds
my_diamonds <- ddply(diamonds, c("cut", "color"), summarise,
my.carat=runif(n=1, min=0.5, max=1))
If I were to do this manually, it would look something like this:
# Use the ecdf for the first entry: cut=="Fair" and color=="D"
my_diamonds$percentile <- NA
my_diamonds$percentile[my_diamonds$cut=="Fair" & my_diamonds$color=="D"] <-
all_ecdfs[["Fair.D"]](my_diamonds$my.carat[my_diamonds$cut=="Fair" & my_diamonds$color=="D"])
Seems like there should be some way to use ldply or lapply to do this automatically, but I can't figure it out.
Here's how I would do it using dplyr to make the ecdfs, and vectorizing to get the values for your data.
#get ecdfs
library(dplyr)
z <- diamonds %>% group_by(cut, color) %>%
summarise(x = list(ecdf(carat)))
Now you have a dataframe z with the functions in a list in column x.
Call the function on our data. We go by row, and get the matching cut and color, then call the function on carat:
z$x[z$cut == my_diamonds$cut & z$color == my_diamonds$color][[1]](my_diamonds$my.carat)

Resources