Create a columns conditioned to two possible levels contained in another column - r

I have the following dataset
out
# A tibble: 1,356 x 7
ID GROUP Gender Age Education tests score
<dbl> <chr> <dbl> <dbl> <dbl> <chr> <dbl>
1 1 TRAINING 1 74 18 ADAS_CogT0 14.7
2 1 TRAINING 1 74 18 ROCF_CT0 32
3 1 TRAINING 1 74 18 ROCF_IT0 3.7
4 1 TRAINING 1 74 18 ROCF_RT0 3.9
5 1 TRAINING 1 74 18 PVF_T0 41.3
6 1 TRAINING 1 74 18 SVF_T0 40
7 1 TRAINING 1 74 18 ADAS_CogT7 16
8 1 TRAINING 1 74 18 ROCF_CT7 33
9 1 TRAINING 1 74 18 ROCF_IT7 1.7
10 1 TRAINING 1 74 18 ROCF_RT7 2.4
If I would like to create a column where in place of the tests ending with T0 would corresponf the value score0 whereas in place of tests ending with T7 the value would be score7`, which are the possible way to fulfill this?

Please be so kind put the data in your posts. >> dput(df)
You could use a combination of case_when and str_detect
library(dplyr)
library(stringr)
df <- structure(
list(
ID = 1:10,
GROUP = rep('TRAINING', 10),
Gender = rep(1, 10),
Education = rep(74, 10),
test = c(
'ADAS_CogT0',
'ROCF_CT0',
'ROCF_IT0',
'ROCF_RT0',
'PVF_T0',
'SVF_T0',
'ADAS_CogT7',
'ROCF_CT7',
'ROCF_IT7',
'ROCF_RT7'
),
score = c(14.7,32,3.7,3.9,41.3,40,16,33,1.7,2.4)
),
row.names = c(1:10),
class = "data.frame"
)
df2 <- df %>%
mutate(new = case_when(str_detect(test, 'T0') ~ 'score0',
str_detect(test, 'T7') ~ 'score7',
TRUE ~ test)
)
ID GROUP Gender Education test score new
1 1 TRAINING 1 74 ADAS_CogT0 14.7 score0
2 2 TRAINING 1 74 ROCF_CT0 32.0 score0
3 3 TRAINING 1 74 ROCF_IT0 3.7 score0
4 4 TRAINING 1 74 ROCF_RT0 3.9 score0
5 5 TRAINING 1 74 PVF_T0 41.3 score0
6 6 TRAINING 1 74 SVF_T0 40.0 score0
7 7 TRAINING 1 74 ADAS_CogT7 16.0 score7
8 8 TRAINING 1 74 ROCF_CT7 33.0 score7
9 9 TRAINING 1 74 ROCF_IT7 1.7 score7
10 10 TRAINING 1 74 ROCF_RT7 2.4 score7

Do you want the output to be string 'score0' and 'score7' ?
You may try -
library(dplyr)
out %>%
mutate(result = case_when(grepl('T0$', tests) ~ 'score0',
grepl('T7$', tests) ~ 'score7'))
# ID GROUP Gender Age Education tests score result
#1 1 TRAINING 1 74 18 ADAS_CogT0 14.7 score0
#2 1 TRAINING 1 74 18 ROCF_CT0 32.0 score0
#3 1 TRAINING 1 74 18 ROCF_IT0 3.7 score0
#4 1 TRAINING 1 74 18 ROCF_RT0 3.9 score0
#5 1 TRAINING 1 74 18 PVF_T0 41.3 score0
#6 1 TRAINING 1 74 18 SVF_T0 40.0 score0
#7 1 TRAINING 1 74 18 ADAS_CogT7 16.0 score7
#8 1 TRAINING 1 74 18 ROCF_CT7 33.0 score7
#9 1 TRAINING 1 74 18 ROCF_IT7 1.7 score7
#10 1 TRAINING 1 74 18 ROCF_RT7 2.4 score7
Or another option with readr::parse_number.
out %>%
mutate(result = paste0('score', readr::parse_number(tests)))

Related

Calculate Percentage by Group with multiple columns in R

I have several data frames with monthly data, I would like to find the percentage distribution for each product and for each month. I have problem with multiple columns with months. Currently, I can only get a percentage by group for one month.
data <- data.frame(group = rep(LETTERS[1:3], each = 4),
Product = letters[1:4],
January = sample(1:100,12),
February = sample(1:100,12))
data_new1 <- transform(data,
perc = ave(January,
group,
FUN = prop.table))
data_new1$perc<-round(data_new1$perc, 2)
> data_new1
group Product January February perc
1 A a 12 16 0.05
2 A b 73 75 0.32
3 A c 78 11 0.34
4 A d 65 35 0.29
5 B a 86 63 0.36
6 B b 33 71 0.14
7 B c 92 49 0.38
8 B d 30 60 0.12
9 C a 91 59 0.37
10 C b 31 45 0.12
11 C c 99 7 0.40
12 C d 28 50 0.11
tidyverse
library(dplyr)
data %>%
group_by(group) %>%
mutate(across(c("January", "February"), proportions, .names = "{.col}_perc")) %>%
ungroup()
# A tibble: 12 x 6
group Product January February January_perc February_perc
<chr> <chr> <int> <int> <dbl> <dbl>
1 A a 49 40 0.426 0.252
2 A b 1 3 0.00870 0.0189
3 A c 19 50 0.165 0.314
4 A d 46 66 0.4 0.415
5 B a 61 82 0.218 0.285
6 B b 88 51 0.314 0.177
7 B c 32 75 0.114 0.260
8 B d 99 80 0.354 0.278
9 C a 6 31 0.0397 0.373
10 C b 8 5 0.0530 0.0602
11 C c 92 20 0.609 0.241
12 C d 45 27 0.298 0.325
base
data <- data.frame(group = rep(LETTERS[1:3], each = 4),
Product = letters[1:4],
January = sample(1:100,12),
February = sample(1:100,12))
tmp <- sapply(c("January", "February"), function (x) ave(data[[x]], data$group, FUN = prop.table))
colnames(tmp) <- paste0(colnames(tmp), "_perc")
res <- cbind(data, tmp)
res
#> group Product January February January_perc February_perc
#> 1 A a 42 73 0.18260870 0.238562092
#> 2 A b 67 92 0.29130435 0.300653595
#> 3 A c 58 90 0.25217391 0.294117647
#> 4 A d 63 51 0.27391304 0.166666667
#> 5 B a 48 15 0.21621622 0.081521739
#> 6 B b 16 82 0.07207207 0.445652174
#> 7 B c 80 75 0.36036036 0.407608696
#> 8 B d 78 12 0.35135135 0.065217391
#> 9 C a 81 16 0.32793522 0.117647059
#> 10 C b 83 81 0.33603239 0.595588235
#> 11 C c 11 1 0.04453441 0.007352941
#> 12 C d 72 38 0.29149798 0.279411765
Created on 2021-12-20 by the reprex package (v2.0.1)
data.table
library(data.table)
COLS <- c("January", "February")
COLS_RES <- paste0(COLS, "_perc")
setDT(data)[, (COLS_RES) := lapply(.SD, proportions), by = group, .SDcol = COLS][]
These calculations are easier if your data is structured in a tidy way. In your case, January and February should probably be one single variable called month or something.
Example:
Underneath, I use tidyr::pivot_longer() to combine January and February into one column. Then I use the package dplyr to group the dataframe and calculate perc. I'm not using prop.table(), but I believe you just want the proportion of observation to the total of that group and month.
library(dplyr)
library(tidyr)
# To make the sampling underneath reproducable
set.seed(1)
data <- data.frame(
group = rep(LETTERS[1:3], each = 4),
Product = letters[1:4],
January = sample(1:100,12),
February = sample(1:100,12)
)
data %>%
pivot_longer(c(January, February), names_to = "month", values_to = "x") %>%
group_by(group, month) %>%
mutate(
perc = round(x/sum(x), 2)
)
I hope this is what you were looking for.
Another dplyr solution:
library(dplyr)
data %>%
group_by(group) %>%
mutate(across(c(2:5),
~./sum(.)*100, .names = "{.col}_pct"))
# A tibble: 12 × 10
# Groups: group [3]
group Product Jan Feb Mar May Jan_pct Feb_pct Mar_pct May_pct
<chr> <chr> <int> <int> <int> <int> <dbl> <dbl> <dbl> <dbl>
1 A a 14 14 95 50 8 18.4 44.4 20.9
2 A b 100 33 28 32 57.1 43.4 13.1 13.4
3 A c 11 16 13 95 6.29 21.1 6.07 39.7
4 A d 50 13 78 62 28.6 17.1 36.4 25.9
5 B a 29 42 72 13 22.0 33.9 20.3 7.07
6 B b 3 4 88 41 2.27 3.23 24.9 22.3
7 B c 30 68 94 86 22.7 54.8 26.6 46.7
8 B d 70 10 100 44 53.0 8.06 28.2 23.9
9 C a 4 88 45 84 3.96 43.6 24.2 30.7
10 C b 52 12 26 55 51.5 5.94 14.0 20.1
11 C c 26 20 23 57 25.7 9.90 12.4 20.8
12 C d 19 82 92 78 18.8 40.6 49.5 28.5
Data:
data <- data.frame(group = rep(LETTERS[1:3], each = 4),
Product = letters[1:4],
Jan = sample(1:100,12),
Feb = sample(1:100,12),
Mar = sample(1:100, 12),
May = sample(1:100, 12))

Looping linear regression output in a data frame in r

I have a dataset below in which I want to do linear regression for each country and state and then cbind the predicted values in the dataset:
Final data frame after adding three more columns:
I have done it for one country and one area but want to do it for each country and area and put the predicted, upper and lower limit values back in the data set by cbind:
data <- data.frame(country = c("US","US","US","US","US","US","US","US","US","US","UK","UK","UK","UK","UK"),
Area = c("G","G","G","G","G","I","I","I","I","I","A","A","A","A","A"),
week = c(1,2,3,4,5,1,2,3,4,5,1,2,3,4,5),amount = c(12,23,34,32,12,12,34,45,65,45,45,34,23,43,43))
data_1 <- data[(data$country=="US" & data$Area=="G"),]
model <- lm(amount ~ week, data = data_1)
pre <- predict(model,newdata = data_1,interval = "prediction",level = 0.95)
pre
How can I loop this for other combination of country and Area?
...and a Base R solution:
data <- data.frame(country = c("US","US","US","US","US","US","US","US","US","US","UK","UK","UK","UK","UK"),
Area = c("G","G","G","G","G","I","I","I","I","I","A","A","A","A","A"),
week = c(1,2,3,4,5,1,2,3,4,5,1,2,3,4,5),amount = c(12,23,34,32,12,12,34,45,65,45,45,34,23,43,43))
splitVar <- paste0(data$country,"-",data$Area)
dfList <- split(data,splitVar)
result <- do.call(rbind,lapply(dfList,function(x){
model <- lm(amount ~ week, data = x)
cbind(x,predict(model,newdata = x,interval = "prediction",level = 0.95))
}))
result
...the results:
country Area week amount fit lwr upr
UK-A.11 UK A 1 45 36.6 -6.0463638 79.24636
UK-A.12 UK A 2 34 37.1 -1.3409128 75.54091
UK-A.13 UK A 3 23 37.6 0.6671656 74.53283
UK-A.14 UK A 4 43 38.1 -0.3409128 76.54091
UK-A.15 UK A 5 43 38.6 -4.0463638 81.24636
US-G.1 US G 1 12 20.8 -27.6791493 69.27915
US-G.2 US G 2 23 21.7 -21.9985147 65.39851
US-G.3 US G 3 34 22.6 -19.3841749 64.58417
US-G.4 US G 4 32 23.5 -20.1985147 67.19851
US-G.5 US G 5 12 24.4 -24.0791493 72.87915
US-I.6 US I 1 12 20.8 -33.8985900 75.49859
US-I.7 US I 2 34 30.5 -18.8046427 79.80464
US-I.8 US I 3 45 40.2 -7.1703685 87.57037
US-I.9 US I 4 65 49.9 0.5953573 99.20464
US-I.10 US I 5 45 59.6 4.9014100 114.29859
We can also use function augment from package broom to get your desired information:
library(purrr)
library(broom)
data %>%
group_by(country, Area) %>%
nest() %>%
mutate(models = map(data, ~ lm(amount ~ week, data = .)),
aug = map(models, ~ augment(.x, interval = "prediction"))) %>%
unnest(aug) %>%
select(country, Area, amount, week, .fitted, .lower, .upper)
# A tibble: 15 x 7
# Groups: country, Area [3]
country Area amount week .fitted .lower .upper
<chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
1 US G 12 1 20.8 -27.7 69.3
2 US G 23 2 21.7 -22.0 65.4
3 US G 34 3 22.6 -19.4 64.6
4 US G 32 4 23.5 -20.2 67.2
5 US G 12 5 24.4 -24.1 72.9
6 US I 12 1 20.8 -33.9 75.5
7 US I 34 2 30.5 -18.8 79.8
8 US I 45 3 40.2 -7.17 87.6
9 US I 65 4 49.9 0.595 99.2
10 US I 45 5 59.6 4.90 114.
11 UK A 45 1 36.6 -6.05 79.2
12 UK A 34 2 37.1 -1.34 75.5
13 UK A 23 3 37.6 0.667 74.5
14 UK A 43 4 38.1 -0.341 76.5
15 UK A 43 5 38.6 -4.05 81.2
Here is a tidyverse way to do this for every combination of country and Area.
library(tidyverse)
data %>%
group_by(country, Area) %>%
nest() %>%
mutate(model = map(data, ~ lm(amount ~ week, data = .x)),
result = map2(model, data, ~data.frame(predict(.x, newdata = .y,
interval = "prediction",level = 0.95)))) %>%
ungroup %>%
select(-model) %>%
unnest(c(data, result))
# country Area week amount fit lwr upr
# <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
# 1 US G 1 12 20.8 -27.7 69.3
# 2 US G 2 23 21.7 -22.0 65.4
# 3 US G 3 34 22.6 -19.4 64.6
# 4 US G 4 32 23.5 -20.2 67.2
# 5 US G 5 12 24.4 -24.1 72.9
# 6 US I 1 12 20.8 -33.9 75.5
# 7 US I 2 34 30.5 -18.8 79.8
# 8 US I 3 45 40.2 -7.17 87.6
# 9 US I 4 65 49.9 0.595 99.2
#10 US I 5 45 59.6 4.90 114.
#11 UK A 1 45 36.6 -6.05 79.2
#12 UK A 2 34 37.1 -1.34 75.5
#13 UK A 3 23 37.6 0.667 74.5
#14 UK A 4 43 38.1 -0.341 76.5
#15 UK A 5 43 38.6 -4.05 81.2
And one more:
library(tidyverse)
data %>%
mutate(CountryArea=paste0(country,Area) %>% factor %>% fct_inorder) %>%
split(.$CountryArea) %>%
map(~lm(amount~week, data=.)) %>%
map(predict, interval = "prediction",level = 0.95) %>%
reduce(rbind) %>%
cbind(data, .)
country Area week amount fit lwr upr
1 US G 1 12 20.8 -27.6791493 69.27915
2 US G 2 23 21.7 -21.9985147 65.39851
3 US G 3 34 22.6 -19.3841749 64.58417
4 US G 4 32 23.5 -20.1985147 67.19851
5 US G 5 12 24.4 -24.0791493 72.87915
6 US I 1 12 20.8 -33.8985900 75.49859
7 US I 2 34 30.5 -18.8046427 79.80464
8 US I 3 45 40.2 -7.1703685 87.57037
9 US I 4 65 49.9 0.5953573 99.20464
10 US I 5 45 59.6 4.9014100 114.29859
11 UK A 1 45 36.6 -6.0463638 79.24636
12 UK A 2 34 37.1 -1.3409128 75.54091
13 UK A 3 23 37.6 0.6671656 74.53283
14 UK A 4 43 38.1 -0.3409128 76.54091
15 UK A 5 43 38.6 -4.0463638 81.24636

Linear regression after select() method in R

I am trying to create a linear regression model from openintro::babies that predicts a baby's birthweight from all other variables in the data except case.
I have to following code:
library(tidyverse)
library(tidymodels)
babies <- openintro::babies %>%
drop_na() %>%
mutate(bwt = 28.3495 * bwt) %>%
mutate(weight = 0.453592 * weight)
linear_reg() %>%
set_engine("lm") %>%
fit(formula = bwt ~ ., data = babies %>% select(-case)) %>%
pluck("fit") %>%
augment(babies)
but in my output, I obtain the case variable as well
# A tibble: 1,174 x 14
case bwt gestation parity age height weight smoke .fitted .resid .hat .sigma .cooksd .std.resid
<int> <dbl> <int> <int> <int> <int> <dbl> <int> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 1 3402. 284 0 27 62 45.4 0 3459. -56.8 0.00374 449. 0.00000863 -0.127
2 2 3203. 282 0 33 64 61.2 0 3547. -344. 0.00227 449. 0.000191 -0.767
3 3 3629. 279 0 28 64 52.2 1 3244. 385. 0.00291 449. 0.000307 0.858
4 5 3062. 282 0 23 67 56.7 1 3396. -334. 0.00475 449. 0.000379 -0.746
5 6 3856. 286 0 25 62 42.2 0 3474. 381. 0.00495 449. 0.000515 0.851
6 7 3912. 244 0 33 62 80.7 0 3065. 848. 0.0137 448. 0.00715 1.90
7 8 3742. 245 0 23 65 63.5 0 3124. 618. 0.00716 449. 0.00197 1.38
8 9 3402. 289 0 25 62 56.7 0 3558. -156. 0.00301 449. 0.0000521 -0.348
9 10 4054. 299 0 30 66 61.7 1 3591. 463. 0.00462 449. 0.000710 1.03
10 11 3969. 351 0 27 68 54.4 0 4527. -558. 0.0221 449. 0.00510 -1.26
# ... with 1,164 more rows
I'm not sure is it the correct way or it is inherent with the output.
Your code is correct. You're getting the case column because of the augment(babies) call, but if you replace it with augment(babies %>% select(-case)) you wont get that column. In other words, the regression model you're fitting does not take into acount the case column].

Purrr Multiply index data frame with dataframe

Thank you all for reading this problem.
What i would like to do is multiply my testdata with my index file while matching columns.
So multiplying Dp_water with Dp_water and iterating over all index vars kcal, fat, prot, carbs.
In my test data i have for 10 individuals data on consumption of 4 food groups in grams.
for each individual i would like to calculate the kcal fat prot carb intake.
For each individual i would like to make a new variable
Dp_water_kcal, Dp_coffee_kcal, Dp_soup_kcal , Dp_soda_kcal
Dp_water_fat, Dp_coffee_fat, Dp_soup_fat , Dp_soda_fat
ect...
library(tidyverse)
Sample data
Index file
index <- data.frame(Variable=c("Dp_water","Dp_coffee","Dp_soup","Dp_soda"),
kcal=c(0,10,20,40),
fat=c(0,5,10,15),
prot=c(2,4,6,8),
carbs=c(3,6,9,12))
index <- index %>%
pivot_longer(c(kcal,fat,prot,carbs)) %>%
pivot_wider(names_from = Variable, values_from = value)
> index
# A tibble: 4 x 5
name Dp_water Dp_coffee Dp_soup Dp_soda
<chr> <dbl> <dbl> <dbl> <dbl>
1 kcal 0 10 20 40
2 fat 0 5 10 15
3 prot 2 4 6 8
4 carbs 3 6 9 12
Below subject data consumption of 4 foodgroups.
test_data <- data.frame(Dp_water=c(11:20),
Dp_coffee=c(31:40),
Dp_soup=c(21:30),
Dp_soda=c(41:50),
id=1:10)
Dp_water Dp_coffee Dp_soup Dp_soda id
1 11 31 21 41 1
2 12 32 22 42 2
3 13 33 23 43 3
4 14 34 24 44 4
5 15 35 25 45 5
6 16 36 26 46 6
7 17 37 27 47 7
8 18 38 28 48 8
9 19 39 29 49 9
10 20 40 30 50 10
If i do the following it works. But i would like to do this for all variables and not only kcal. And i would like to be able to keep the id column.
test_data %>%
select(-id) %>%
map2_dfr(., test_data[match(names(.), names(test_data))], ~.x/100 * .y) %>%
set_names(paste0(names(.), "_kcal"))
# A tibble: 10 x 4
Dp_water_kcal Dp_coffee_kcal Dp_soup_kcal Dp_soda_kcal
<dbl> <dbl> <dbl> <dbl>
1 1.21 9.61 4.41 16.8
2 1.44 10.2 4.84 17.6
3 1.69 10.9 5.29 18.5
4 1.96 11.6 5.76 19.4
5 2.25 12.2 6.25 20.2
6 2.56 13.0 6.76 21.2
7 2.89 13.7 7.29 22.1
8 3.24 14.4 7.84 23.0
9 3.61 15.2 8.41 24.0
10 4 16 9 25
Thank you all for any help!

How can I transform multiple repeated measures from wide to long format?

I have a data set that looks like that:
id <- c(1:3)
gender <- factor(c("male","female","female"))
age <- c(51,69,44)
cortisol_1 <- c(23,32,54)
cortisol_2 <- c(34,52,49)
cortisol_3 <- c(34,65,12)
blood_1 <- c(12,64,54)
blood_2 <- c(52,32,75)
blood_3 <- c(12,12,75)
temp_1 <- c(38.5,38.7,37.9)
temp_3 <- c(36.5,36.4,37.1)
df <- data.frame(id,gender,age,cortisol_1,cortisol_2,cortisol_3,blood_1,blood_2,blood_3,temp_1,temp_3)
df
id gender age cortisol_1 cortisol_2 cortisol_3 blood_1 blood_2 blood_3 temp_1 temp_3
1 1 male 51 23 34 34 12 52 12 38.5 36.5
2 2 female 69 32 52 65 64 32 12 38.7 36.4
3 3 female 44 54 49 12 54 75 75 37.9 37.1
So I have cortisol level and blood pressure which were measured annually at three time points. However, body temperature was only assessed at baseline and wave 3.
How can I change the data structure from wide to long? I would hope that the data looks like that:
id gender wave cortisol blood temp
1 1 male 1 23 12 38.5
2 1 male 2 34 52 NA
3 1 male 3 34 12 36.5
4 2 female 1 32 64 37.7
5 2 female 2 52 32 NA
6 2 female 3 65 12 36.4
7 3 female 1 54 54 37.9
8 3 female 2 49 75 NA
9 3 female 3 12 75 37.1
Best
Jascha
We can use pivot_longer
library(dplyr)
library(tidyr)
df %>%
pivot_longer(cols = -c(id, gender, age),
names_to = c('.value', 'grp'), names_sep = "_") %>%
select(-grp)
-output
# A tibble: 9 x 6
# id gender age cortisol blood temp
# <int> <fct> <dbl> <dbl> <dbl> <dbl>
#1 1 male 51 23 12 38.5
#2 1 male 51 34 52 NA
#3 1 male 51 34 12 36.5
#4 2 female 69 32 64 38.7
#5 2 female 69 52 32 NA
#6 2 female 69 65 12 36.4
#7 3 female 44 54 54 37.9
#8 3 female 44 49 75 NA
#9 3 female 44 12 75 37.1

Resources