Amit Levinson 2022-10-24
A repository containing answers for questions in forums.
You might find discrepancies between the number of times requested in the question and my response. This is mainly for ease of printing.
How can I iterate over every column in my dataframe and calculate a t.test between half of the elemnts (e.g. 1:50 vs 51:100) (10/24/2022)
library(purrr)
library(broom)
dat <- data.frame(
col1 = sample(c(1:100), size = 100, replace = T),
col2 = sample(c(1:100), size = 100, replace = T),
col3 = sample(c(1:100), size = 100, replace = T)
)
calculate_ttest <- function(column) {
t.test(column[50:100], column[1:49], alternative= "less", paired=FALSE, mu=0, conf.level = 0.9, var.equal = TRUE) |>
broom::tidy()
}
dat %>%
map_dfr(calculate_ttest, .id = 'columnname')## # A tibble: 3 × 11
## columnname estimate estimate1 estimate2 statistic p.value parameter conf.low
## <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 col1 4.70 54.8 50.1 0.752 0.773 98 -Inf
## 2 col2 -9.74 41.7 51.4 -1.58 0.0591 98 -Inf
## 3 col3 -7.94 45.6 53.5 -1.38 0.0852 98 -Inf
## # … with 3 more variables: conf.high <dbl>, method <chr>, alternative <chr>
How can we generate \~50 formulas to pass onto a model from a given set of variables? \*\*
library(dplyr)##
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
##
## filter, lag
## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, union
library(purrr)
library(broom)
mtcars_vars <- names(mtcars)
# Generate possible formulas
mtcars_formulas <- expand.grid(mtcars_vars,mtcars_vars,mtcars_vars) %>%
filter(Var1 != Var2, Var2 != Var3, Var1 != Var3) %>%
sample_n(10)
# Function to create formulas
f <- function (Var1, Var2, Var3) paste(Var1, paste(Var2, Var3, sep = " + "), sep = " ~ ")
formulas <- mtcars_formulas %>%
pmap_chr(f)
head(formulas)## [1] "drat ~ hp + cyl" "qsec ~ am + mpg" "wt ~ carb + vs"
## [4] "am ~ disp + carb" "drat ~ gear + mpg" "vs ~ hp + cyl"
# Iterate across the formulas and tidy the output with formula as name
models <- map(formulas, ~ lm(., data = mtcars)) %>%
set_names(formulas) %>%
map_dfr(tidy, .id = ".x")
# Output
models## # A tibble: 30 × 6
## .x term estimate std.error statistic p.value
## <chr> <chr> <dbl> <dbl> <dbl> <dbl>
## 1 drat ~ hp + cyl (Intercept) 5.07 0.257 19.8 2.30e-18
## 2 drat ~ hp + cyl hp 0.00340 0.00176 1.94 6.26e- 2
## 3 drat ~ hp + cyl cyl -0.318 0.0674 -4.72 5.53e- 5
## 4 qsec ~ am + mpg (Intercept) 13.8 0.853 16.1 5.10e-16
## 5 qsec ~ am + mpg am -2.69 0.566 -4.76 5.00e- 5
## 6 qsec ~ am + mpg mpg 0.258 0.0468 5.50 6.27e- 6
## 7 wt ~ carb + vs (Intercept) 3.33 0.450 7.39 3.86e- 8
## 8 wt ~ carb + vs carb 0.100 0.112 0.890 3.81e- 1
## 9 wt ~ carb + vs vs -0.895 0.360 -2.49 1.90e- 2
## 10 am ~ disp + carb (Intercept) 0.782 0.163 4.80 4.47e- 5
## # … with 20 more rows
The categories on the X-axis are too condensed and overflow, can we expand them to more than one row?
library(ggplot2)
# Fake data
dat <- data.frame(
category = sample(c(stringr::sentences[1:3], stringr::words[1:4]), size = 100, replace = T)
)
# Option 1 - Category to y axis instead of splitting
p1 <- ggplot(dat) +
geom_bar(aes(y = category)) +
labs(title = "Put the category on the Y-axis")
# Option 2: Stringr::str_wrap
p2 <- ggplot(dat) +
geom_bar(aes(x = stringr::str_wrap(category, 15))) +
labs(title = "use stringr:str_wrap(variable, 15)")I have X classes (groups) and would like to run a t-test for groups within each class separately
library(ggplot2)
library(dplyr)
library(broom)
mtcars %>%
group_by(gear) %>%
summarize(t.test(mpg , cyl, data = cur_data()) %>% tidy())## # A tibble: 3 × 11
## gear estimate estimate1 estimate2 statistic p.value parameter conf.low
## <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 3 8.64 16.1 7.47 9.36 0.0000000326 17.4 6.70
## 2 4 19.9 24.5 4.67 12.8 0.0000000286 11.8 16.5
## 3 5 15.4 21.4 6 4.95 0.00504 4.72 7.24
## # … with 3 more variables: conf.high <dbl>, method <chr>, alternative <chr>
# Using Purrr & unnest
library(purrr)
library(tidyr)
mtcars %>%
group_nest(gear) %>%
summarise(ttest = map(data, ~ (t.test(x = .$mpg , y = .$cyl, data = .x)))) %>%
mutate(ttest = map(ttest, tidy)) %>%
unnest(ttest)## # A tibble: 3 × 10
## estimate estimate1 estimate2 statistic p.value parameter conf.low conf.high
## <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 8.64 16.1 7.47 9.36 3.26e-8 17.4 6.70 10.6
## 2 19.9 24.5 4.67 12.8 2.86e-8 11.8 16.5 23.3
## 3 15.4 21.4 6 4.95 5.04e-3 4.72 7.24 23.5
## # … with 2 more variables: method <chr>, alternative <chr>
How can you randomly sample 100 values with replacement from a vector and calculate an IQR, 10000 times?
# Fake data
vec <- 1:100
# Option 1: Functional
calc_iqr <- function() {
inner_vec <- sample(vec, size = 100, replace = T)
IQR(inner_vec)
}
values_a <- replicate(100, expr = calc_iqr())
# Output ------------------------------------------------------------------
values_a## [1] 50.25 46.00 50.25 43.25 47.50 57.00 48.00 52.25 49.00 48.25 52.00 39.25
## [13] 42.50 62.25 51.50 51.75 56.50 41.75 53.25 50.00 43.00 61.25 51.50 45.00
## [25] 44.25 53.25 53.25 43.00 51.00 50.50 52.25 46.75 47.25 46.50 47.00 50.75
## [37] 52.50 60.00 44.25 49.50 54.25 45.75 51.00 47.50 46.75 52.25 55.75 49.00
## [49] 54.50 49.00 49.75 46.00 48.25 52.50 51.25 51.50 40.50 48.50 42.25 53.50
## [61] 45.50 48.25 47.75 63.75 52.00 37.75 50.25 52.50 44.25 49.75 44.50 48.00
## [73] 46.50 45.50 43.25 40.25 62.00 49.25 46.00 48.25 51.50 46.50 51.25 57.50
## [85] 54.50 50.00 47.25 44.75 44.00 46.25 40.25 46.75 50.00 47.00 42.00 38.25
## [97] 44.00 50.25 40.25 51.50
# Option 2: For loop
values_b <- vector(mode = "numeric", length = 100)
output <- for (i in 1:100) {
inner_vec <- sample(vec, size = 100, replace = T)
values_b[[i]] <- IQR(inner_vec)
}
# Output ------------------------------------------------------------------
values_b## [1] 44.25 53.25 48.25 45.00 51.50 48.25 54.00 49.50 57.50 44.50 54.50 54.75
## [13] 47.25 45.25 47.25 49.50 46.75 40.25 44.25 52.25 53.25 48.50 41.25 46.00
## [25] 60.00 53.25 48.50 47.50 51.50 50.75 46.25 45.00 58.25 51.25 44.25 46.50
## [37] 40.00 51.00 40.50 46.50 53.00 44.75 49.25 41.00 50.50 49.75 46.50 52.25
## [49] 54.50 49.50 56.00 47.25 41.50 51.00 55.00 49.25 56.25 42.25 50.00 44.50
## [61] 61.25 44.25 48.25 56.25 55.00 50.75 48.50 57.50 50.50 48.25 58.75 50.25
## [73] 50.25 53.50 47.00 45.75 48.25 45.50 44.00 52.25 48.00 44.00 44.50 52.25
## [85] 53.50 51.50 35.25 45.50 47.00 40.25 47.25 54.50 48.25 46.25 58.25 53.25
## [97] 54.00 48.50 57.50 49.00