I want to take a df as the one below and want to cut/bin/group/sample into groups of size=20.
Ideally, this "binning" occurs randomly across IDs rather then consecutively from top row to bottom row).
E.g. IDs 2, 29 and 71 have counts of 7,7,6 and would fit nicely into a "bin" of size=20.
I want to achieve the minimum number of bins and do not care about order of IDs (the more random they are, the better).
set.seed(123)
df <- data.frame(
ID = as.numeric(1:100),
Count = as.numeric(sample(1:8, size = 100, replace = T)))
Desired outcome would be a dataframe/tibble looking something like the below with optimum random sampling and minimising bin number.
Bin_size=20 is the parameter set by me (the ideal outcome is exact 20 (=20) but <20 is ok, however, >20 is not ok).
Each Bin should be given a number (e.g. if I have 10 bins, I would like them to be called Bin_number 1-10).
ID, Count, Bin_size, Bin_number
ID 2, 7, 20, 1
ID 29, 7, 20, 1
ID 71, 6, 20, 1
etc.
Where 7+7+6 = 20 (etc.)
Any help with this would be much appreciated. I have been wondering about cumsum and group_by but could not figure it out.
if you need more details, I'm happy to provide them.
thanks!
The BBmisc package has a simple (though not optimized) bin packing algorithm that might be useful:
library(BBmisc)
library(dplyr)
df %>%
as_tibble() %>%
mutate(bin = binPack(Count, 20),
bin_size = ave(Count, bin, FUN = sum)) %>%
arrange(bin)
# A tibble: 100 x 4
ID Count bin bin_size
<dbl> <dbl> <int> <dbl>
1 11 4 1 20
2 17 8 1 20
3 27 8 1 20
4 22 4 2 20
5 42 8 2 20
6 56 8 2 20
7 34 4 3 20
8 62 8 3 20
9 79 8 3 20
10 40 4 4 20
# ... with 90 more rows
Do you want something like this,
df<-df%>%arrange(Count)%>%mutate(Sum=cumsum(Count),Bin_size=20)
df<-df%>%mutate(Bin_number=cut(Sum,breaks = seq(0,c(max(Sum)+20),20),labels = F,right = T))
I have a tibble which has stored variables taken at different points in the sea and at different depths, but I need to condense all the depths of the same point into a single row following a specific formula (the summation of the sum of values X and X+1 times the subtraction of the depth of X+1 minus depth of X and so on...), which I have wrote on excel as a way of better explaining what I'm trying to do
And here is an small sample of the (edited) data I'm working with
long lat station depth no3
<dbl> <dbl> <dbl> <dbl> <dbl>
1 -71.1 -32 1 0 9
2 -71.1 -32 1 5 14
3 -71.1 -32 1 10 10
4 -71.1 -32 1 20 11
5 -71.6 -32 2 0 13
6 -71.6 -32 2 5 8
7 -71.6 -32 2 10 2
8 -71.6 -32 2 20 6
9 -71.6 -32 2 50 4
10 -71.6 -32 2 75 9
# ... with 942 more rows
From what I read here in similar questions, I could use aggregate or merge but those only do the summation, and I don't know how to get it to do the entire equation. I'll appreciate any suggestion, I'm new to R and if I haven't been very clear (or the solution is actually quite simple) I'm sorry
You can use the lead function to create the sums based on the following row's data (use lag for the previous row), and then sum this new column in a summarize:
df <- data.frame(
depth = c(0, 5, 10, 20, 50),
NO3 = c(3, 5, 6, 2, 3)) %>%
mutate(a = (lead(NO3) + NO3)*(lead(depth) - depth))
df
depth NO3 a
1 0 3 40
2 5 5 55
3 10 6 80
4 20 2 150
5 50 3 NA
df %>%
summarize(b = sum(a, na.rm = TRUE))
b
1 325
Note that the na.rm in the sum is key here since the lead functions create NA values in the final row. These can be filled using the default argument.
EDIT:
If you'd like to apply this to more than just one column, you can use the "scoped" variants of mutate and summarize, by adding _at or _if to the end of these functions.
df2 <- data.frame(
depth = c(0, 5, 10, 20, 50),
NO3 = c(3, 5, 6, 2, 3),
NO4 = c(1, 2, 3, 4, 5),
NO5 = c(5, 4, 3, 2, 1))
_at functions require either a names vector or an index vector to determine which columns to operate on. All three of these will return the same thing, where .x refers to the column being modified:
df2 %>%
mutate_at(c("NO3", "NO4", "NO5"), ~(lead(.x) + .x)*(lead(depth)-depth)) %>%
summarize_at(c("NO3", "NO4", "NO5"), sum, na.rm = T)
df2 %>%
mutate_at(vars(NO3:NO5), ~(lead(.x) + .x)*(lead(depth)-depth)) %>%
summarize_at(vars(NO3:NO5), sum, na.rm = T)
df2 %>%
mutate_at(2:4, ~(lead(.x) + .x)*(lead(depth)-depth)) %>%
summarize_at(2:4, sum, na.rm = T)
NO3 NO4 NO5
1 325 380 220
_if functions need a "predicate function" that determines whether a column will be operated on. Either of these, which check the name of the column, would work:
df2 %>%
mutate_if(str_detect(colnames(.), "NO"), ~(lead(.x) + .x)*(lead(depth)-depth)) %>%
summarize_if(str_detect(colnames(.), "NO"), sum, na.rm = T)
df2 %>%
mutate_if(!str_detect(colnames(.), "depth"), ~(lead(.x) + .x)*(lead(depth)-depth)) %>%
summarize_if(!str_detect(colnames(.), "depth"), sum, na.rm = T)
NO3 NO4 NO5
1 325 380 220
I'm in the process of trying out a dplyr-based workflow (rather than using mostly data.table, which I'm used to), and I've come across a problem that I can't find an equivalent dplyr solution to. I commonly run into the scenario where I need to conditionally update/replace several columns based on a single condition. Here's some example code, with my data.table solution:
library(data.table)
# Create some sample data
set.seed(1)
dt <- data.table(site = sample(1:6, 50, replace=T),
space = sample(1:4, 50, replace=T),
measure = sample(c('cfl', 'led', 'linear', 'exit'), 50,
replace=T),
qty = round(runif(50) * 30),
qty.exit = 0,
delta.watts = sample(10.5:100.5, 50, replace=T),
cf = runif(50))
# Replace the values of several columns for rows where measure is "exit"
dt <- dt[measure == 'exit',
`:=`(qty.exit = qty,
cf = 0,
delta.watts = 13)]
Is there a simple dplyr solution to this same problem? I'd like to avoid using ifelse because I don't want to have to type the condition multiple times - this is a simplified example, but there are sometimes many assignments based on a single condition.
These solutions (1) maintain the pipeline, (2) do not overwrite the input and (3) only require that the condition be specified once:
1a) mutate_cond Create a simple function for data frames or data tables that can be incorporated into pipelines. This function is like mutate but only acts on the rows satisfying the condition:
mutate_cond <- function(.data, condition, ..., envir = parent.frame()) {
condition <- eval(substitute(condition), .data, envir)
.data[condition, ] <- .data[condition, ] %>% mutate(...)
.data
}
DF %>% mutate_cond(measure == 'exit', qty.exit = qty, cf = 0, delta.watts = 13)
1b) mutate_last This is an alternative function for data frames or data tables which again is like mutate but is only used within group_by (as in the example below) and only operates on the last group rather than every group. Note that TRUE > FALSE so if group_by specifies a condition then mutate_last will only operate on rows satisfying that condition.
mutate_last <- function(.data, ...) {
n <- n_groups(.data)
indices <- attr(.data, "indices")[[n]] + 1
.data[indices, ] <- .data[indices, ] %>% mutate(...)
.data
}
DF %>%
group_by(is.exit = measure == 'exit') %>%
mutate_last(qty.exit = qty, cf = 0, delta.watts = 13) %>%
ungroup() %>%
select(-is.exit)
2) factor out condition Factor out the condition by making it an extra column which is later removed. Then use ifelse, replace or arithmetic with logicals as illustrated. This also works for data tables.
library(dplyr)
DF %>% mutate(is.exit = measure == 'exit',
qty.exit = ifelse(is.exit, qty, qty.exit),
cf = (!is.exit) * cf,
delta.watts = replace(delta.watts, is.exit, 13)) %>%
select(-is.exit)
3) sqldf We could use SQL update via the sqldf package in the pipeline for data frames (but not data tables unless we convert them -- this may represent a bug in dplyr. See dplyr issue 1579). It may seem that we are undesirably modifying the input in this code due to the existence of the update but in fact the update is acting on a copy of the input in the temporarily generated database and not on the actual input.
library(sqldf)
DF %>%
do(sqldf(c("update '.'
set 'qty.exit' = qty, cf = 0, 'delta.watts' = 13
where measure = 'exit'",
"select * from '.'")))
4) row_case_when Also check out row_case_when defined in
Returning a tibble: how to vectorize with case_when? . It uses a syntax similar to case_when but applies to rows.
library(dplyr)
DF %>%
row_case_when(
measure == "exit" ~ data.frame(qty.exit = qty, cf = 0, delta.watts = 13),
TRUE ~ data.frame(qty.exit, cf, delta.watts)
)
Note 1: We used this as DF
set.seed(1)
DF <- data.frame(site = sample(1:6, 50, replace=T),
space = sample(1:4, 50, replace=T),
measure = sample(c('cfl', 'led', 'linear', 'exit'), 50,
replace=T),
qty = round(runif(50) * 30),
qty.exit = 0,
delta.watts = sample(10.5:100.5, 50, replace=T),
cf = runif(50))
Note 2: The problem of how to easily specify updating a subset of rows is also discussed in dplyr issues 134, 631, 1518 and 1573 with 631 being the main thread and 1573 being a review of the answers here.
You can do this with magrittr's two-way pipe %<>%:
library(dplyr)
library(magrittr)
dt[dt$measure=="exit",] %<>% mutate(qty.exit = qty,
cf = 0,
delta.watts = 13)
This reduces the amount of typing, but is still much slower than data.table.
Here's a solution I like:
mutate_when <- function(data, ...) {
dots <- eval(substitute(alist(...)))
for (i in seq(1, length(dots), by = 2)) {
condition <- eval(dots[[i]], envir = data)
mutations <- eval(dots[[i + 1]], envir = data[condition, , drop = FALSE])
data[condition, names(mutations)] <- mutations
}
data
}
It lets you write things like e.g.
mtcars %>% mutate_when(
mpg > 22, list(cyl = 100),
disp == 160, list(cyl = 200)
)
which is quite readable -- although it may not be as performant as it could be.
As eipi10 shows above, there's not a simple way to do a subset replacement in dplyr because DT uses pass-by-reference semantics vs dplyr using pass-by-value. dplyr requires the use of ifelse() on the whole vector, whereas DT will do the subset and update by reference (returning the whole DT). So, for this exercise, DT will be substantially faster.
You could alternatively subset first, then update, and finally recombine:
dt.sub <- dt[dt$measure == "exit",] %>%
mutate(qty.exit= qty, cf= 0, delta.watts= 13)
dt.new <- rbind(dt.sub, dt[dt$measure != "exit",])
But DT is gonna be substantially faster:
(editted to use eipi10's new answer)
library(data.table)
library(dplyr)
library(microbenchmark)
microbenchmark(dt= {dt <- dt[measure == 'exit',
`:=`(qty.exit = qty,
cf = 0,
delta.watts = 13)]},
eipi10= {dt[dt$measure=="exit",] %<>% mutate(qty.exit = qty,
cf = 0,
delta.watts = 13)},
alex= {dt.sub <- dt[dt$measure == "exit",] %>%
mutate(qty.exit= qty, cf= 0, delta.watts= 13)
dt.new <- rbind(dt.sub, dt[dt$measure != "exit",])})
Unit: microseconds
expr min lq mean median uq max neval cld
dt 591.480 672.2565 747.0771 743.341 780.973 1837.539 100 a
eipi10 3481.212 3677.1685 4008.0314 3796.909 3936.796 6857.509 100 b
alex 3412.029 3637.6350 3867.0649 3726.204 3936.985 5424.427 100 b
I just stumbled across this and really like mutate_cond() by #G. Grothendieck, but thought it might come in handy to also handle new variables. So, below has two additions:
Unrelated: Second last line made a bit more dplyr by using filter()
Three new lines at the beginning get variable names for use in mutate(), and initializes any new variables in the data frame before mutate() occurs. New variables are initialized for the remainder of the data.frame using new_init, which is set to missing (NA) as a default.
mutate_cond <- function(.data, condition, ..., new_init = NA, envir = parent.frame()) {
# Initialize any new variables as new_init
new_vars <- substitute(list(...))[-1]
new_vars %<>% sapply(deparse) %>% names %>% setdiff(names(.data))
.data[, new_vars] <- new_init
condition <- eval(substitute(condition), .data, envir)
.data[condition, ] <- .data %>% filter(condition) %>% mutate(...)
.data
}
Here are some examples using the iris data:
Change Petal.Length to 88 where Species == "setosa". This will work in the original function as well as this new version.
iris %>% mutate_cond(Species == "setosa", Petal.Length = 88)
Same as above, but also create a new variable x (NA in rows not included in the condition). Not possible before.
iris %>% mutate_cond(Species == "setosa", Petal.Length = 88, x = TRUE)
Same as above, but rows not included in the condition for x are set to FALSE.
iris %>% mutate_cond(Species == "setosa", Petal.Length = 88, x = TRUE, new_init = FALSE)
This example shows how new_init can be set to a list to initialize multiple new variables with different values. Here, two new variables are created with excluded rows being initialized using different values (x initialised as FALSE, y as NA)
iris %>% mutate_cond(Species == "setosa" & Sepal.Length < 5,
x = TRUE, y = Sepal.Length ^ 2,
new_init = list(FALSE, NA))
One concise solution would be to do the mutation on the filtered subset and then add back the non-exit rows of the table:
library(dplyr)
dt %>%
filter(measure == 'exit') %>%
mutate(qty.exit = qty, cf = 0, delta.watts = 13) %>%
rbind(dt %>% filter(measure != 'exit'))
You could split the dataset and do a regular mutate call on the TRUE part.
the split can be done with either dplyr::group_split() or base::split(), I like the base version better here since it preserves names, see the discussion at https://github.com/tidyverse/dplyr/issues/4223 .
df1 <- data.frame(site = sample(1:6, 50, replace=T),
space = sample(1:4, 50, replace=T),
measure = sample(c('cfl', 'led', 'linear', 'exit'), 50,
replace=T),
qty = round(runif(50) * 30),
qty.exit = 0,
delta.watts = sample(10.5:100.5, 50, replace=T),
cf = runif(50),
stringsAsFactors = F)
library(tidyverse)
df1 %>%
group_split(measure == "exit", .keep = FALSE) %>%
modify_at(2, ~mutate(.,qty.exit = qty, cf = 0, delta.watts = 13)) %>%
bind_rows()
#> # A tibble: 50 × 7
#> site space measure qty qty.exit delta.watts cf
#> <int> <int> <chr> <dbl> <dbl> <dbl> <dbl>
#> 1 5 1 linear 22 0 100. 0.126
#> 2 3 3 led 12 0 61.5 0.161
#> 3 6 1 led 26 0 25.5 0.307
#> 4 5 2 cfl 16 0 26.5 0.865
#> 5 6 3 linear 19 0 57.5 0.684
#> 6 1 4 led 12 0 14.5 0.802
#> 7 6 4 led 5 0 90.5 0.547
#> 8 5 4 linear 28 0 54.5 0.171
#> 9 1 2 linear 5 0 24.5 0.775
#> 10 1 2 cfl 24 0 96.5 0.144
#> # … with 40 more rows
df1 %>%
split(~measure == "exit") %>%
modify_at("TRUE", ~mutate(.,qty.exit = qty, cf = 0, delta.watts = 13)) %>%
bind_rows()
#> site space measure qty qty.exit delta.watts cf
#> 1 5 1 linear 22 0 100.5 0.125646491
#> 2 3 3 led 12 0 61.5 0.160692291
#> 3 6 1 led 26 0 25.5 0.307239765
#> 4 5 2 cfl 16 0 26.5 0.864969074
#> 5 6 3 linear 19 0 57.5 0.683945200
#> 6 1 4 led 12 0 14.5 0.802398642
#> 7 6 4 led 5 0 90.5 0.547211378
#> 8 5 4 linear 28 0 54.5 0.170614207
#> 9 1 2 linear 5 0 24.5 0.774603932
#> 10 1 2 cfl 24 0 96.5 0.144310557
#> 11 3 4 linear 21 0 93.5 0.682622390
#> 12 4 4 led 2 0 48.5 0.941718646
#> 13 4 4 cfl 2 0 100.5 0.918448627
#> 14 5 2 led 11 0 63.5 0.998143780
#> 15 4 1 led 21 0 53.5 0.644740176
#> 16 1 3 cfl 5 0 28.5 0.110610285
#> 17 1 3 linear 24 0 41.5 0.538868200
#> 18 4 3 led 29 0 19.5 0.998474289
#> 19 2 3 cfl 4 0 22.5 0.008167536
#> 20 5 1 led 20 0 56.5 0.740833476
#> 21 3 2 led 5 0 44.5 0.223967706
#> 22 1 4 led 27 0 32.5 0.199850583
#> 23 3 4 cfl 17 0 61.5 0.104023080
#> 24 1 3 cfl 11 0 34.5 0.399036247
#> 25 2 3 linear 29 0 65.5 0.600678235
#> 26 2 4 cfl 23 0 29.5 0.291611352
#> 27 6 2 linear 13 0 37.5 0.225021614
#> 28 2 3 led 17 0 62.5 0.879606956
#> 29 2 4 led 29 0 51.5 0.301759669
#> 30 5 1 led 11 0 54.5 0.793816856
#> 31 2 3 led 20 0 29.5 0.514759195
#> 32 3 4 linear 6 0 68.5 0.475085443
#> 33 1 4 led 21 0 34.5 0.133207588
#> 34 2 4 linear 25 0 80.5 0.164279355
#> 35 5 3 led 7 0 73.5 0.252937836
#> 36 6 2 led 15 0 99.5 0.554864929
#> 37 3 2 linear 6 0 44.5 0.377257874
#> 38 4 4 exit 15 15 13.0 0.000000000
#> 39 3 3 exit 10 10 13.0 0.000000000
#> 40 5 1 exit 15 15 13.0 0.000000000
#> 41 4 2 exit 1 1 13.0 0.000000000
#> 42 5 3 exit 10 10 13.0 0.000000000
#> 43 1 3 exit 14 14 13.0 0.000000000
#> 44 5 2 exit 12 12 13.0 0.000000000
#> 45 2 2 exit 30 30 13.0 0.000000000
#> 46 6 3 exit 28 28 13.0 0.000000000
#> 47 1 1 exit 14 14 13.0 0.000000000
#> 48 3 3 exit 21 21 13.0 0.000000000
#> 49 4 2 exit 13 13 13.0 0.000000000
#> 50 4 3 exit 12 12 13.0 0.000000000
Created on 2022-10-07 by the reprex package (v2.0.1)
mutate_cond is a great function, but it gives an error if there is an NA in the column(s) used to create the condition. I feel that a conditional mutate should simply leave such rows alone. This matches the behavior of filter(), which returns rows when the condition is TRUE, but omits both rows with FALSE and NA.
With this small change the function works like a charm:
mutate_cond <- function(.data, condition, ..., envir = parent.frame()) {
condition <- eval(substitute(condition), .data, envir)
condition[is.na(condition)] = FALSE
.data[condition, ] <- .data[condition, ] %>% mutate(...)
.data
}
I don't actually see any changes to dplyr that would make this much easier. case_when is great for when there are multiple different conditions and outcomes for one column but it doesn't help for this case where you want to change multiple columns based on one condition. Similarly, recode saves typing if you are replacing multiple different values in one column but doesn't help with doing so in multiple columns at once. Finally, mutate_at etc. only apply conditions to the column names not the rows in the dataframe. You could potentially write a function for mutate_at that would do it but I can't figure out how you would make it behave differently for different columns.
That said here is how I would approach it using nest form tidyr and map from purrr.
library(data.table)
library(dplyr)
library(tidyr)
library(purrr)
# Create some sample data
set.seed(1)
dt <- data.table(site = sample(1:6, 50, replace=T),
space = sample(1:4, 50, replace=T),
measure = sample(c('cfl', 'led', 'linear', 'exit'), 50,
replace=T),
qty = round(runif(50) * 30),
qty.exit = 0,
delta.watts = sample(10.5:100.5, 50, replace=T),
cf = runif(50))
dt2 <- dt %>%
nest(-measure) %>%
mutate(data = if_else(
measure == "exit",
map(data, function(x) mutate(x, qty.exit = qty, cf = 0, delta.watts = 13)),
data
)) %>%
unnest()
With the creation of rlang, a slightly modified version of Grothendieck's 1a example is possible, eliminating the need for the envir argument, as enquo() captures the environment that .p is created in automatically.
mutate_rows <- function(.data, .p, ...) {
.p <- rlang::enquo(.p)
.p_lgl <- rlang::eval_tidy(.p, .data)
.data[.p_lgl, ] <- .data[.p_lgl, ] %>% mutate(...)
.data
}
dt %>% mutate_rows(measure == "exit", qty.exit = qty, cf = 0, delta.watts = 13)
I think this answer has not been mentioned before. It runs almost as fast as the 'default' data.table-solution..
Use base::replace()
df %>% mutate( qty.exit = replace( qty.exit, measure == 'exit', qty[ measure == 'exit'] ),
cf = replace( cf, measure == 'exit', 0 ),
delta.watts = replace( delta.watts, measure == 'exit', 13 ) )
replace recycles the replacement value, so when you want the values of columns qty entered into colums qty.exit, you have to subset qty as well... hence the qty[ measure == 'exit'] in the first replacement..
now, you will probably not want to retype the measure == 'exit' all the time... so you can create an index-vector containing that selection, and use it in the functions above.
#build an index-vector matching the condition
index.v <- which( df$measure == 'exit' )
df %>% mutate( qty.exit = replace( qty.exit, index.v, qty[ index.v] ),
cf = replace( cf, index.v, 0 ),
delta.watts = replace( delta.watts, index.v, 13 ) )
benchmarks
# Unit: milliseconds
# expr min lq mean median uq max neval
# data.table 1.005018 1.053370 1.137456 1.112871 1.186228 1.690996 100
# wimpel 1.061052 1.079128 1.218183 1.105037 1.137272 7.390613 100
# wimpel.index 1.043881 1.064818 1.131675 1.085304 1.108502 4.192995 100
At the expense of breaking with the usual dplyr syntax, you can use within from base:
dt %>% within(qty.exit[measure == 'exit'] <- qty[measure == 'exit'],
delta.watts[measure == 'exit'] <- 13)
It seems to integrate well with the pipe, and you can do pretty much anything you want inside it.
I have a question that I find kind of hard to explain with a MRE and in an easy
way to answer, mostly because I don't fully understand where the problem lies
myself. So that's my sorry for being vague preamble.
I have a tibble with many sample and reference measurements, for which I want
to do some linear interpolation for each sample. I do this now by taking out
all the reference measurements, rescaling them to sample measurements using
approx, and then patching it back in. But because I take it out first, I
cannot do it nicely in a group_by dplyr pipe way. right now I do it with a
really ugly workaround where I add empty (NA) newly created columns to the
sample tibble, then do it with a for-loop.
So my question is really: how can I implement the approx part within groups
into the pipe, so that I can do everything within groups? I've experimented
with dplyr::do(), and ran into the vignette on "programming with dplyr", but
searching mostly gives me broom::augment and lm stuff that I think operates
differently... (e.g. see
Using approx() with groups in dplyr). This thread also seems promising: How do you use approx() inside of mutate_at()?
Somebody on irc recommended using a conditional mutate, with case_when, but I
don't fully understand where and how within this context yet.
I think the problem lies in the fact that I want to filter out part of the data
for the following mutate operations, but the mutate operations rely on the
grouped data that I just filtered out, if that makes any sense.
Here's a MWE:
library(tidyverse) # or just dplyr, tibble
# create fake data
data <- data.frame(
# in reality a dttm with the measurement time
timestamp = c(rep("a", 7), rep("b", 7), rep("c", 7)),
# measurement cycle, normally 40 for sample, 41 for reference
cycle = rep(c(rep(1:3, 2), 4), 3),
# wheather the measurement is a reference or a sample
isref = rep(c(rep(FALSE, 3), rep(TRUE, 4)), 3),
# measurement intensity for mass 44
r44 = c(28:26, 30:26, 36, 33, 31, 38, 34, 33, 31, 18, 16, 15, 19, 18, 17)) %>%
# measurement intensity for mass 45, normally also masses up to mass 49
mutate(r45 = r44 + rnorm(21, 20))
# of course this could be tidied up to "intensity" with a new column "mass"
# (44, 45, ...), but that would make making comparisons even harder...
# overview plot
data %>%
ggplot(aes(x = cycle, y = r44, colour = isref)) +
geom_line() +
geom_line(aes(y = r45), linetype = 2) +
geom_point() +
geom_point(aes(y = r45), shape = 1) +
facet_grid(~ timestamp)
# what I would like to do
data %>%
group_by(timestamp) %>%
do(target_cycle = approx(x = data %>% filter(isref) %>% pull(r44),
y = data %>% filter(isref) %>% pull(cycle),
xout = data %>% filter(!isref) %>% pull(r44))$y) %>%
unnest()
# immediately append this new column to the original dataframe for all the
# samples (!isref) and then apply another approx for those values.
# here's my current attempt for one of the timestamps
matchref <- function(dat) {
# split the data into sample gas and reference gas
ref <- filter(dat, isref)
smp <- filter(dat, !isref)
# calculate the "target cycle", the points at which the reference intensity
# 44 matches the sample intensity 44 with linear interpolation
target_cycle <- approx(x = ref$r44,
y = ref$cycle, xout = smp$r44)
# append the target cycle to the sample gas
smp <- smp %>%
group_by(timestamp) %>%
mutate(target = target_cycle$y)
# linearly interpolate each reference gas to the target cycle
ref <- ref %>%
group_by(timestamp) %>%
# this is needed because the reference has one more cycle
mutate(target = c(target_cycle$y, NA)) %>%
# filter out all the failed ones (no interpolation possible)
filter(!is.na(target)) %>%
# calculate interpolated value based on r44 interpolation (i.e., don't
# actually interpolate this value but shift it based on the 44
# interpolation)
mutate(r44 = approx(x = cycle, y = r44, xout = target)$y,
r45 = approx(x = cycle, y = r45, xout = target)$y) %>%
select(timestamp, target, r44:r45)
# add new reference gas intensities to the correct sample gasses by the target cycle
left_join(smp, ref, by = c("time", "target"))
}
matchref(data)
# and because now "target" must be length 3 (the group size) or one, not 9
# I have to create this ugly for-loop
# for which I create a copy of data that has the new columns to be created
mr <- data %>%
# filter the sample gasses (since we convert ref to sample)
filter(!isref) %>%
# add empty new columns
mutate(target = NA, r44 = NA, r45 = NA)
# apply matchref for each group timestamp
for (grp in unique(data$timestamp)) {
mr[mr$timestamp == grp, ] <- matchref(data %>% filter(timestamp == grp))
}
Here's one approach that spreads the references and samples to new columns. I drop r45 for simplicity in this example.
data %>%
select(-r45) %>%
mutate(isref = ifelse(isref, "REF", "SAMP")) %>%
spread(isref, r44) %>%
group_by(timestamp) %>%
mutate(target_cycle = approx(x = REF, y = cycle, xout = SAMP)$y) %>%
ungroup
gives,
# timestamp cycle REF SAMP target_cycle
# <fct> <dbl> <dbl> <dbl> <dbl>
# 1 a 1 30 28 3
# 2 a 2 29 27 4
# 3 a 3 28 26 NA
# 4 a 4 27 NA NA
# 5 b 1 31 26 NA
# 6 b 2 38 36 2.5
# 7 b 3 34 33 4
# 8 b 4 33 NA NA
# 9 c 1 15 31 NA
# 10 c 2 19 18 3
# 11 c 3 18 16 2.5
# 12 c 4 17 NA NA
Edit to address comment below
To retain r45 you can use a gather-unite-spread approach like this:
df %>%
mutate(isref = ifelse(isref, "REF", "SAMP")) %>%
gather(r, value, r44:r45) %>%
unite(ru, r, isref, sep = "_") %>%
spread(ru, value) %>%
group_by(timestamp) %>%
mutate(target_cycle_r44 = approx(x = r44_REF, y = cycle, xout = r44_SAMP)$y) %>%
ungroup
giving,
# # A tibble: 12 x 7
# timestamp cycle r44_REF r44_SAMP r45_REF r45_SAMP target_cycle_r44
# <fct> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
# 1 a 1 30 28 49.5 47.2 3
# 2 a 2 29 27 48.8 48.7 4
# 3 a 3 28 26 47.2 46.8 NA
# 4 a 4 27 NA 47.9 NA NA
# 5 b 1 31 26 51.4 45.7 NA
# 6 b 2 38 36 57.5 55.9 2.5
# 7 b 3 34 33 54.3 52.4 4
# 8 b 4 33 NA 52.0 NA NA
# 9 c 1 15 31 36.0 51.7 NA
# 10 c 2 19 18 39.1 37.9 3
# 11 c 3 18 16 39.2 35.3 2.5
# 12 c 4 17 NA 39.0 NA NA