# Load tidyverse package
library(tidyverse)
# Load a package that can read excel files
library(readxl)
library(readr)Presentation 2: Tidyverse
Want to code along?
Load packages
Check working directory
Check working directory so you know from where you work.
getwd()[1] "/Users/srz223/Desktop/DataLab/FromExceltoR/Teachers/Presentations"Importing data
Often we will work with large datasets that already exist in i.e. an excel sheet or a tab separated file (.tsv). We can easily load that data into R, either with the read_excel function or by clicking on ‘Import Dataset’ in the Environment tab (right). We can also load data in via a command. Let’s do this now. Navigate to the data from your working directory. Use the tap-button to check what your options are.
crohns <- read_excel("../../Data/crohns_disease.xlsx")A first look at the data
Print first few lines of your dataset
head(crohns) # A tibble: 6 × 9
ID nrAdvE BMI height country sex age weight treat
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr>
1 19908 4 25.2 163 c1 F 47 67 placebo
2 19909 4 23.8 164 c1 F 53 64 d1
3 19910 1 23.0 164 c1 F 68 62 placebo
4 20908 1 25.7 165 c1 F 48 70 d2
5 20909 2 26.0 170 c1 F 67 75 placebo
6 20910 2 28.7 168 c1 F 54 81 d1 Get the dimension of your dataset
dim(crohns)[1] 117 9How many observations (rows) do have?
nrow(crohns)[1] 117How many data columns are there and what are their types? Both ‘str’ and ‘summary’ will you what column types you have. Summary has some extra summary stats on numeric columns.
summary(crohns) ID nrAdvE BMI height
Min. :19908 Min. : 0.000 Min. :16.00 Min. :124.0
1st Qu.:23909 1st Qu.: 0.000 1st Qu.:23.05 1st Qu.:157.0
Median :25919 Median : 1.000 Median :25.15 Median :162.0
Mean :34103 Mean : 2.034 Mean :26.06 Mean :162.7
3rd Qu.:51909 3rd Qu.: 3.000 3rd Qu.:28.40 3rd Qu.:166.0
Max. :54937 Max. :12.000 Max. :44.06 Max. :182.0
country sex age weight
Length:117 Length:117 Min. :19.00 Min. : 36.00
Class :character Class :character 1st Qu.:48.00 1st Qu.: 59.00
Mode :character Mode :character Median :56.00 Median : 68.00
Mean :54.66 Mean : 69.03
3rd Qu.:62.00 3rd Qu.: 76.00
Max. :75.00 Max. :117.00
treat
Length:117
Class :character
Mode :character
str(crohns)tibble [117 × 9] (S3: tbl_df/tbl/data.frame)
$ ID : num [1:117] 19908 19909 19910 20908 20909 ...
$ nrAdvE : num [1:117] 4 4 1 1 2 2 3 0 1 0 ...
$ BMI : num [1:117] 25.2 23.8 23.1 25.7 25.9 ...
$ height : num [1:117] 163 164 164 165 170 168 161 168 154 157 ...
$ country: chr [1:117] "c1" "c1" "c1" "c1" ...
$ sex : chr [1:117] "F" "F" "F" "F" ...
$ age : num [1:117] 47 53 68 48 67 54 53 53 47 58 ...
$ weight : num [1:117] 67 64 62 70 75 81 69 74 76 82 ...
$ treat : chr [1:117] "placebo" "d1" "placebo" "d2" ...The anatomy of tidyverse
Tidyverse is a collection of R packages that are great for data wrangling and visualizations. Data wrangling with functions from the Tidyverse are often used with a specific syntax:
The name of the variable you are creating. Can omit if you don’t want to save the result.
The name of the dataset we are working on.
The function you want to apply on the dataset (and whatever arguments must be provided to the function).
In tidyverse we use the pipe symbol %>% to chain multiple functions together. The term pipe comes from the fact that we pipe the output from one function into another function as the input.
It is a good idea to make a new line after each pipe symbol.
# new_object <- dataset %>%
# function1(arguments...) %>%
# function2(arguments...)Count, distinct, sort
Count and distinct are very useful to get information about your dataset!
Variables (columns) can be numeric or categorical (characters, factors). Use the str() function to see the structure of your dataset.
crohns %>%
str()tibble [117 × 9] (S3: tbl_df/tbl/data.frame)
$ ID : num [1:117] 19908 19909 19910 20908 20909 ...
$ nrAdvE : num [1:117] 4 4 1 1 2 2 3 0 1 0 ...
$ BMI : num [1:117] 25.2 23.8 23.1 25.7 25.9 ...
$ height : num [1:117] 163 164 164 165 170 168 161 168 154 157 ...
$ country: chr [1:117] "c1" "c1" "c1" "c1" ...
$ sex : chr [1:117] "F" "F" "F" "F" ...
$ age : num [1:117] 47 53 68 48 67 54 53 53 47 58 ...
$ weight : num [1:117] 67 64 62 70 75 81 69 74 76 82 ...
$ treat : chr [1:117] "placebo" "d1" "placebo" "d2" ...distinct() tells us how many different levels a categorical variable has.
# How many different treatments do we have?
crohns %>%
distinct(treat)# A tibble: 3 × 1
treat
<chr>
1 placebo
2 d1
3 d2 #From how many different countries do we have data?
crohns %>%
distinct(country)# A tibble: 2 × 1
country
<chr>
1 c1
2 c2 count() does tabulation of categorical variables. Total number of lines, i.e. patients in the current dataset. Observe, this matches with the number of lines you can see in the Environment tab.
crohns %>%
count()# A tibble: 1 × 1
n
<int>
1 117# How many lines, i.e. patients do we have per treatment?
crohns %>%
count(treat)# A tibble: 3 × 2
treat n
<chr> <int>
1 d1 39
2 d2 39
3 placebo 39# Is our dataset balanced?
# How many patients do we have for each age?
crohns %>%
count(age)# A tibble: 43 × 2
age n
<dbl> <int>
1 19 1
2 28 1
3 29 1
4 30 1
5 33 1
6 35 1
7 36 1
8 38 1
9 39 3
10 40 2
# ℹ 33 more rows# Perhaps this is more useful: How many patients are older than 65?
crohns %>%
count(age > 65)# A tibble: 2 × 2
`age > 65` n
<lgl> <int>
1 FALSE 96
2 TRUE 21Note we haven’t saved anything here, we just get output to the console sorted in a certain way. This helps us to check if the data looks correct and get an impression.
Filtering data (selecting rows) with filter()
How we subset dataset into subsets we find interesting. For example only female patients:
crohns %>%
filter(sex == 'F') # processed from left to right# A tibble: 100 × 9
ID nrAdvE BMI height country sex age weight treat
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr>
1 19908 4 25.2 163 c1 F 47 67 placebo
2 19909 4 23.8 164 c1 F 53 64 d1
3 19910 1 23.0 164 c1 F 68 62 placebo
4 20908 1 25.7 165 c1 F 48 70 d2
5 20909 2 26.0 170 c1 F 67 75 placebo
6 20910 2 28.7 168 c1 F 54 81 d1
7 21908 3 26.6 161 c1 F 53 69 d1
8 21909 0 26.2 168 c1 F 53 74 placebo
9 21910 1 32.0 154 c1 F 47 76 d2
10 21911 0 33.3 157 c1 F 58 82 placebo
# ℹ 90 more rowsA great about tidyverse: write code the way you think. You always filter by defining conditions. If the condition evaluates to ‘TRUE’ the line is included. See only data lines for patients over 65:
crohns %>%
filter(age > 65)# A tibble: 21 × 9
ID nrAdvE BMI height country sex age weight treat
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr>
1 19910 1 23.0 164 c1 F 68 62 placebo
2 20909 2 26.0 170 c1 F 67 75 placebo
3 22908 5 18.2 159 c1 F 66 46 d2
4 22909 8 30.8 156 c1 M 71 75 placebo
5 22911 3 24.8 182 c1 M 68 82 d2
6 22914 3 25.2 157 c1 F 75 62 d2
7 24912 0 25.2 162 c1 F 69 66 placebo
8 25920 1 27.9 164 c1 F 69 75 d1
9 26908 0 25.8 150 c1 F 73 58 d1
10 26910 0 19.1 165 c1 F 66 52 d1
# ℹ 11 more rowsFrom the above commands we are getting the result printed to the console. This is useful to check something. To save the result, we need to re-assign:
seniors <- crohns %>%
filter(age > 65)View newly created data frame:
seniors# A tibble: 21 × 9
ID nrAdvE BMI height country sex age weight treat
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr>
1 19910 1 23.0 164 c1 F 68 62 placebo
2 20909 2 26.0 170 c1 F 67 75 placebo
3 22908 5 18.2 159 c1 F 66 46 d2
4 22909 8 30.8 156 c1 M 71 75 placebo
5 22911 3 24.8 182 c1 M 68 82 d2
6 22914 3 25.2 157 c1 F 75 62 d2
7 24912 0 25.2 162 c1 F 69 66 placebo
8 25920 1 27.9 164 c1 F 69 75 d1
9 26908 0 25.8 150 c1 F 73 58 d1
10 26910 0 19.1 165 c1 F 66 52 d1
# ℹ 11 more rowsDo we still have all three treatment groups in our subset?
seniors %>%
count(treat)# A tibble: 3 × 2
treat n
<chr> <int>
1 d1 6
2 d2 5
3 placebo 10The world of conditional operators
Now we get lines that fit certain conditions but what if I want to filter on more than one condition? Enter conditional operators!
The ‘and’ operator: &
We can also subset on several conditions. Here are younger patients who received drug 1:
crohns %>%
filter(age <= 65 & treat == 'd1')# A tibble: 33 × 9
ID nrAdvE BMI height country sex age weight treat
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr>
1 19909 4 23.8 164 c1 F 53 64 d1
2 20910 2 28.7 168 c1 F 54 81 d1
3 21908 3 26.6 161 c1 F 53 69 d1
4 21916 0 23.9 177 c1 M 56 75 d1
5 22916 2 30.9 163 c1 F 53 82 d1
6 23908 0 30.4 158 c1 F 55 76 d1
7 23909 0 23.4 156 c1 F 44 57 d1
8 23910 0 26.7 156 c1 F 59 65 d1
9 23912 0 16 150 c1 F 28 36 d1
10 24909 0 22.5 155 c1 F 52 54 d1
# ℹ 23 more rowsThe ‘or’ operator: |
Get patients that were treated with either drug 1 or the placebo:
crohns %>%
filter(treat == 'placebo' | treat == 'd1')# A tibble: 78 × 9
ID nrAdvE BMI height country sex age weight treat
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr>
1 19908 4 25.2 163 c1 F 47 67 placebo
2 19909 4 23.8 164 c1 F 53 64 d1
3 19910 1 23.0 164 c1 F 68 62 placebo
4 20909 2 26.0 170 c1 F 67 75 placebo
5 20910 2 28.7 168 c1 F 54 81 d1
6 21908 3 26.6 161 c1 F 53 69 d1
7 21909 0 26.2 168 c1 F 53 74 placebo
8 21911 0 33.3 157 c1 F 58 82 placebo
9 21914 6 28.4 170 c1 M 58 82 placebo
10 21916 0 23.9 177 c1 M 56 75 d1
# ℹ 68 more rowsThe ‘not’ operator: !
crohns %>%
filter(treat != 'placebo')# A tibble: 78 × 9
ID nrAdvE BMI height country sex age weight treat
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr>
1 19909 4 23.8 164 c1 F 53 64 d1
2 20908 1 25.7 165 c1 F 48 70 d2
3 20910 2 28.7 168 c1 F 54 81 d1
4 21908 3 26.6 161 c1 F 53 69 d1
5 21910 1 32.0 154 c1 F 47 76 d2
6 21912 5 32.5 152 c1 F 63 75 d2
7 21913 2 37.6 159 c1 F 54 95 d2
8 21915 0 23.0 160 c1 F 54 59 d2
9 21916 0 23.9 177 c1 M 56 75 d1
10 21917 0 36.4 164 c1 F 51 98 d2
# ℹ 68 more rowsOther conditional operators can be found in the first presentation or when querying the function (or just google it).
?dplyr::filterYou can also allow multiple arguments in a variable. Here are the young patients who got treatment with either drug 1 or 2:
crohns %>%
filter(age <= 65 & treat %in% c("d1","d2"))# A tibble: 67 × 9
ID nrAdvE BMI height country sex age weight treat
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr>
1 19909 4 23.8 164 c1 F 53 64 d1
2 20908 1 25.7 165 c1 F 48 70 d2
3 20910 2 28.7 168 c1 F 54 81 d1
4 21908 3 26.6 161 c1 F 53 69 d1
5 21910 1 32.0 154 c1 F 47 76 d2
6 21912 5 32.5 152 c1 F 63 75 d2
7 21913 2 37.6 159 c1 F 54 95 d2
8 21915 0 23.0 160 c1 F 54 59 d2
9 21916 0 23.9 177 c1 M 56 75 d1
10 21917 0 36.4 164 c1 F 51 98 d2
# ℹ 57 more rowsSelecting variables (columns) with select()
We can choose to only include certain columns. Here, we select only BMI, age and the number of adverse events:
crohns %>%
select(nrAdvE, BMI, age)# A tibble: 117 × 3
nrAdvE BMI age
<dbl> <dbl> <dbl>
1 4 25.2 47
2 4 23.8 53
3 1 23.0 68
4 1 25.7 48
5 2 26.0 67
6 2 28.7 54
7 3 26.6 53
8 0 26.2 53
9 1 32.0 47
10 0 33.3 58
# ℹ 107 more rowsWe can also make a negative selection that excludes the named column(s). The ID doesn’t give us any information since the data is anonymized:
without_id <- crohns %>%
select(-ID)We have saved the dataset without the ID column in at new variable. Let’s have a look at this:
without_id# A tibble: 117 × 8
nrAdvE BMI height country sex age weight treat
<dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr>
1 4 25.2 163 c1 F 47 67 placebo
2 4 23.8 164 c1 F 53 64 d1
3 1 23.0 164 c1 F 68 62 placebo
4 1 25.7 165 c1 F 48 70 d2
5 2 26.0 170 c1 F 67 75 placebo
6 2 28.7 168 c1 F 54 81 d1
7 3 26.6 161 c1 F 53 69 d1
8 0 26.2 168 c1 F 53 74 placebo
9 1 32.0 154 c1 F 47 76 d2
10 0 33.3 157 c1 F 58 82 placebo
# ℹ 107 more rowsTransformation of data with mutate()
We can create new columns based on other columns with the mutate() function.
This is our original tibble:
crohns# A tibble: 117 × 9
ID nrAdvE BMI height country sex age weight treat
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr>
1 19908 4 25.2 163 c1 F 47 67 placebo
2 19909 4 23.8 164 c1 F 53 64 d1
3 19910 1 23.0 164 c1 F 68 62 placebo
4 20908 1 25.7 165 c1 F 48 70 d2
5 20909 2 26.0 170 c1 F 67 75 placebo
6 20910 2 28.7 168 c1 F 54 81 d1
7 21908 3 26.6 161 c1 F 53 69 d1
8 21909 0 26.2 168 c1 F 53 74 placebo
9 21910 1 32.0 154 c1 F 47 76 d2
10 21911 0 33.3 157 c1 F 58 82 placebo
# ℹ 107 more rowsWe want to add height in meters in a new column. It is important to reassign the dataframe if you want to save the new column.
crohns <- crohns %>%
mutate(height_m = height/100)
crohns# A tibble: 117 × 10
ID nrAdvE BMI height country sex age weight treat height_m
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr> <dbl>
1 19908 4 25.2 163 c1 F 47 67 placebo 1.63
2 19909 4 23.8 164 c1 F 53 64 d1 1.64
3 19910 1 23.0 164 c1 F 68 62 placebo 1.64
4 20908 1 25.7 165 c1 F 48 70 d2 1.65
5 20909 2 26.0 170 c1 F 67 75 placebo 1.7
6 20910 2 28.7 168 c1 F 54 81 d1 1.68
7 21908 3 26.6 161 c1 F 53 69 d1 1.61
8 21909 0 26.2 168 c1 F 53 74 placebo 1.68
9 21910 1 32.0 154 c1 F 47 76 d2 1.54
10 21911 0 33.3 157 c1 F 58 82 placebo 1.57
# ℹ 107 more rowsWe can also create columns based on TRUE/FALSE conditions. According to the CDC, a person with a BMI < 18.5 is underweight:
crohns <- crohns %>%
mutate(underweight = ifelse(BMI < 18.5, "Yes", "No"))
crohns# A tibble: 117 × 11
ID nrAdvE BMI height country sex age weight treat height_m
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr> <dbl>
1 19908 4 25.2 163 c1 F 47 67 placebo 1.63
2 19909 4 23.8 164 c1 F 53 64 d1 1.64
3 19910 1 23.0 164 c1 F 68 62 placebo 1.64
4 20908 1 25.7 165 c1 F 48 70 d2 1.65
5 20909 2 26.0 170 c1 F 67 75 placebo 1.7
6 20910 2 28.7 168 c1 F 54 81 d1 1.68
7 21908 3 26.6 161 c1 F 53 69 d1 1.61
8 21909 0 26.2 168 c1 F 53 74 placebo 1.68
9 21910 1 32.0 154 c1 F 47 76 d2 1.54
10 21911 0 33.3 157 c1 F 58 82 placebo 1.57
# ℹ 107 more rows
# ℹ 1 more variable: underweight <chr>How many patients are underweight?
crohns %>%
count(underweight)# A tibble: 2 × 2
underweight n
<chr> <int>
1 No 113
2 Yes 4Have a look at the mutate() function:
?mutateSummary statistics, revisited with summarize()
Methods from before:
mean(crohns$age)[1] 54.65812max(crohns$age)[1] 75summary(crohns) ID nrAdvE BMI height
Min. :19908 Min. : 0.000 Min. :16.00 Min. :124.0
1st Qu.:23909 1st Qu.: 0.000 1st Qu.:23.05 1st Qu.:157.0
Median :25919 Median : 1.000 Median :25.15 Median :162.0
Mean :34103 Mean : 2.034 Mean :26.06 Mean :162.7
3rd Qu.:51909 3rd Qu.: 3.000 3rd Qu.:28.40 3rd Qu.:166.0
Max. :54937 Max. :12.000 Max. :44.06 Max. :182.0
country sex age weight
Length:117 Length:117 Min. :19.00 Min. : 36.00
Class :character Class :character 1st Qu.:48.00 1st Qu.: 59.00
Mode :character Mode :character Median :56.00 Median : 68.00
Mean :54.66 Mean : 69.03
3rd Qu.:62.00 3rd Qu.: 76.00
Max. :75.00 Max. :117.00
treat height_m underweight
Length:117 Min. :1.240 Length:117
Class :character 1st Qu.:1.570 Class :character
Mode :character Median :1.620 Mode :character
Mean :1.627
3rd Qu.:1.660
Max. :1.820 The summarize() function does the same but in a tidyverse way and gives the result in a table which you can export and send to your colleagues.
crohns %>%
summarize(mean(age),
max(age))# A tibble: 1 × 2
`mean(age)` `max(age)`
<dbl> <dbl>
1 54.7 75We can also specify names for the new columns:
crohns %>%
summarize(mean_age = mean(age),
max_age = max(age))# A tibble: 1 × 2
mean_age max_age
<dbl> <dbl>
1 54.7 75What kind of things can you summarize? Have a look at the help by typing ?summarize into the console, or ‘summarize’ into the help panel and scroll down to ‘Useful functions’.
A useful summarize function is n() which counts the number of lines.
crohns %>%
summarize(mean_age = mean(age),
max_age = max(age),
number_lines = n())# A tibble: 1 × 3
mean_age max_age number_lines
<dbl> <dbl> <int>
1 54.7 75 117Note that R is tolerant of BE/AE spelling differences. summarise() and summarize() are the same function, likewise with color and colour.
Grouping with group_by()
The function group_by() imposes a grouping on a tibble. Group according to sex:
crohns %>%
group_by(sex)# A tibble: 117 × 11
# Groups: sex [2]
ID nrAdvE BMI height country sex age weight treat height_m
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr> <dbl>
1 19908 4 25.2 163 c1 F 47 67 placebo 1.63
2 19909 4 23.8 164 c1 F 53 64 d1 1.64
3 19910 1 23.0 164 c1 F 68 62 placebo 1.64
4 20908 1 25.7 165 c1 F 48 70 d2 1.65
5 20909 2 26.0 170 c1 F 67 75 placebo 1.7
6 20910 2 28.7 168 c1 F 54 81 d1 1.68
7 21908 3 26.6 161 c1 F 53 69 d1 1.61
8 21909 0 26.2 168 c1 F 53 74 placebo 1.68
9 21910 1 32.0 154 c1 F 47 76 d2 1.54
10 21911 0 33.3 157 c1 F 58 82 placebo 1.57
# ℹ 107 more rows
# ℹ 1 more variable: underweight <chr>We can also group according to several variables. How many groups will we get?
crohns %>%
group_by(sex, treat)# A tibble: 117 × 11
# Groups: sex, treat [6]
ID nrAdvE BMI height country sex age weight treat height_m
<dbl> <dbl> <dbl> <dbl> <chr> <chr> <dbl> <dbl> <chr> <dbl>
1 19908 4 25.2 163 c1 F 47 67 placebo 1.63
2 19909 4 23.8 164 c1 F 53 64 d1 1.64
3 19910 1 23.0 164 c1 F 68 62 placebo 1.64
4 20908 1 25.7 165 c1 F 48 70 d2 1.65
5 20909 2 26.0 170 c1 F 67 75 placebo 1.7
6 20910 2 28.7 168 c1 F 54 81 d1 1.68
7 21908 3 26.6 161 c1 F 53 69 d1 1.61
8 21909 0 26.2 168 c1 F 53 74 placebo 1.68
9 21910 1 32.0 154 c1 F 47 76 d2 1.54
10 21911 0 33.3 157 c1 F 58 82 placebo 1.57
# ℹ 107 more rows
# ℹ 1 more variable: underweight <chr>By itself, group_by does nothing, we still get the same dataset returned. But it is very useful in combination with other commands! The reason we want to do it this is way is that we can first impose grouping with group_by() and then pipe, %>%, the resulting tibble into summarize which will respect our grouping. So smart!
crohns %>% # the dataset
group_by(sex) %>% # grouped by sex
summarise(avg = mean(age), # calculate mean of the age
med = median(age), # calc median
stdev = sd(age), # calc standard dev.
n = n()) # get the number of observations# A tibble: 2 × 5
sex avg med stdev n
<chr> <dbl> <dbl> <dbl> <int>
1 F 54.7 55 10.8 100
2 M 54.3 56 10.4 17Now we see why n() is useful: It tells us how many lines, i.e. patients are in each group.
Group by sex and treatment, and calculate stats for the number of adverse events.
crohns %>% # the dataset
group_by(sex, treat) %>% # grouped by sex
summarise(avg = mean(nrAdvE), # calculate mean number of adverse events
med = median(nrAdvE), # calc median
max = max(nrAdvE), # calc max
stdev = sd(nrAdvE), # calc standard dev.
total_events = sum(nrAdvE), # calc cumulative sum
n = n()) # get the number of observations# A tibble: 6 × 8
# Groups: sex [2]
sex treat avg med max stdev total_events n
<chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <int>
1 F d1 1.5 0 7 2.08 51 34
2 F d2 2.12 1 9 2.71 72 34
3 F placebo 2.16 1 12 3.09 69 32
4 M d1 2 0 9 3.94 10 5
5 M d2 2.2 2 6 2.49 11 5
6 M placebo 3.57 3 8 3.41 25 7The might of the pipe operator: %>%
Many commands can be combined with the pipe operator to pipe data through an analysis workflow.
crohns %>% # the dataset
filter(age > 65) %>% # filtered to only people over 65
group_by(sex, treat) %>% # Grouping
summarise(avg = mean(nrAdvE), # calculate mean number of adverse events
med = median(nrAdvE), # calc median
max = max(nrAdvE), # calc max
stdev = sd(nrAdvE), # calc standard dev.
total_events = sum(nrAdvE), # calc cumulative sum
n = n()) %>% # get the number of observations
arrange(avg) # Sort output by the mean# A tibble: 5 × 8
# Groups: sex [2]
sex treat avg med max stdev total_events n
<chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <int>
1 F d1 1.83 0.5 6 2.56 11 6
2 F placebo 2.22 1 12 3.73 20 9
3 F d2 3 3 5 2 9 3
4 M d2 4.5 4.5 6 2.12 9 2
5 M placebo 8 8 8 NA 8 1What if I want to do the same analysis but with only obese patients? The CDC lists a BMI of > 30 as obese.