Tidyverse: Difference between revisions
(→split) |
|||
Line 1,338: | Line 1,338: | ||
<li>[https://www.brodrigues.co/blog/2022-05-26-safer_programs/ Some learnings from functional programming you can use to write safer programs] </li> | <li>[https://www.brodrigues.co/blog/2022-05-26-safer_programs/ Some learnings from functional programming you can use to write safer programs] </li> | ||
</ul> | </ul> | ||
== .x symbol == | |||
[https://community.rstudio.com/t/function-argument-naming-conventions-x-vs-x/7764/2 Function argument naming conventions (`.x` vs `x`)]. Se [https://purrr.tidyverse.org/reference/map.html purrr::map] | |||
== negate() == | == negate() == |
Revision as of 19:43, 6 July 2022
Tidyverse
Import | | readr, readxl | haven, DBI, httr +----- Visualize ------+ | | ggplot2, ggvis | | | | Tidy ------------- Transform tibble dplyr Model tidyr | broom +------ Model ---------+
Cheat sheet
The cheat sheets are downloaded from RStudio
- Data Transformation with dply
- Data Import
- Data Import with readr, tibble, and tidyr (not in RStudio anymore?)
Online
- TidyverseSkeptic by Norm Matloff
- R for Data Science and tidyverse package (it is a collection of ggplot2, tibble, tidyr, readr, purrr, dplyr, stringr & forcats 8 packages).
- tidyverse, among others, was used at Mining CRAN DESCRIPTION Files (tbl_df(), %>%, summarise(), count(), mutate(), arrange(), unite(), ggplot(), filter(), select(), ...). Note that there is a problem to reproduce the result. I need to run cran <- cran[, -14] to remove the MD5sum column.
- Compile R for Data Science to a PDF
- The tidyverse style guide by Hadley Wickham
- Data Wrangling with dplyr and tidyr Cheat Sheet
- Data Wrangling with Tidyverse from the Harvard Chan School of Public Health.
- Best packages for data manipulation in R. It demonstrates to perform the same tasks using data.table and dplyr packages. data.table is faster and it may be a go-to package when performance and memory are the constraints.
- DATA MANIPULATION IN R by Alboukadel Kassambara
- subset data frame columns: pull() [return a vector], select() [return data frame], select_if(), other helper functions
- subset (filter) data frame rows: slice(), filter(), filter_all(), filter_if(), filter_at(), sample_n(), top_n()
- identify and remove duplicate rows: duplicated(), unique(), distinct(). distinct() will keep only distinct variables if variables are specified (cf select() which keeps all rows0.
- ordering rows: arrange(), desc()
- cf stats::reorder() to change a factor variable's order based on another variable. So the output is still a vector. It is useful in creating multiple boxplots. On the other hand, arrange() is to change the row order of a data frame and its input is a data frame.
- desc() can be used in arrange() [see ?desc] and reorder() [see ordered barplot ] too.
- desc(x) is just doing the negative operation -x.
- renaming and adding columns: rename()
- compute and add new variables to a data frame: mutate(), transmutate()
- computing summary statistics (pay to view)
- Data manipulation in r using data frames - an extensive article of basics
- The A to Z of tidyverse from Deeply Trivial
- Summer Institute in Statistics for Big Data (SISBID), SISBID 2020 Modules
- Complete tutorial
Animation to explain
- tidyexplain - Tidy Animated Verbs
- Tidy Data Tutor helps you visualize data analysis pipelines
stringr vs base R
Examples
A Gentle Introduction to Tidy Statistics in R
A Gentle Introduction to Tidy Statistics in R by Thomas Mock on RStudio webinar. Good coverage with step-by-step explanation. See part 1 & part 2 about the data and markdown document. All documents are available in github repository.
Task | R code | Graph |
---|---|---|
Load the libraries | library(tidyverse) library(readxl) library(broom) library(knitr) |
|
Read Excel file | raw_df <- readxl::read_xlsx("ad_treatment.xlsx") dplyr::glimpse(raw_df) |
|
Check distribution | g2 <- ggplot(raw_df, aes(x = age)) + geom_density(fill = "blue") g2 raw_df %>% summarize(min = min(age), max = max(age)) |
File:Check dist.svg |
Data cleaning | raw_df %>% summarize(na_count = sum(is.na(mmse))) |
|
Experimental variables
levels |
# check Ns and levels for our variables table(raw_df$drug_treatment, raw_df$health_status) table(raw_df$drug_treatment, raw_df$health_status, raw_df$sex) # tidy way of looking at variables raw_df %>% group_by(drug_treatment, health_status, sex) %>% count() |
|
Visual Exploratory
Data Analysis |
ggplot(data = raw_df, # add the data aes(x = drug_treatment, y = mmse, # set x, y coordinates color = drug_treatment)) + # color by treatment geom_boxplot() + facet_grid(~health_status) |
File:Onefacet.svg |
Summary Statistics | raw_df %>% glimpse() sum_df <- raw_df %>% mutate( sex = factor(sex, labels = c("Male", "Female")), drug_treatment = factor(drug_treatment, levels = c("Placebo", "Low dose", "High Dose")), health_status = factor(health_status, levels = c("Healthy", "Alzheimer's")) ) %>% group_by(sex, health_status, drug_treatment # group by categorical variables ) %>% summarize( mmse_mean = mean(mmse), # calc mean mmse_se = sd(mmse)/sqrt(n()) # calc standard error ) %>% ungroup() # ungrouping variable is a good habit to prevent errors kable(sum_df) write.csv(sum_df, "adx37_sum_stats.csv") |
|
Plotting summary
statistics |
g <- ggplot(data = sum_df, # add the data aes(x = drug_treatment, #set x, y coordinates y = mmse_mean, group = drug_treatment, # group by treatment color = drug_treatment)) + # color by treatment geom_point(size = 3) + # set size of the dots facet_grid(sex~health_status) # create facets by sex and status g |
File:Twofacets.svg |
ANOVA | # set up the statistics df stats_df <- raw_df %>% # start with data mutate(drug_treatment = factor(drug_treatment, levels = c("Placebo", "Low dose", "High Dose")), sex = factor(sex, labels = c("Male", "Female")), health_status = factor(health_status, levels = c("Healthy", "Alzheimer's"))) glimpse(stats_df) # this gives main effects AND interactions ad_aov <- aov(mmse ~ sex * drug_treatment * health_status, data = stats_df) summary(ad_aov) # this extracts ANOVA output into a nice tidy dataframe tidy_ad_aov <- tidy(ad_aov) # which we can save to Excel write.csv(tidy_ad_aov, "ad_aov.csv") |
|
Post-hocs | # pairwise t.tests ad_pairwise <- pairwise.t.test(stats_df$mmse, stats_df$sex:stats_df$drug_treatment:stats_df$health_status, p.adj = "none") # look at the posthoc p.values in a tidy dataframe kable(head(tidy(ad_pairwise))) # call and tidy the tukey posthoc tidy_ad_tukey <- tidy( TukeyHSD(ad_aov, which = 'sex:drug_treatment:health_status')) |
|
Publication plot | sig_df <- tribble( ~drug_treatment, ~ health_status, ~sex, ~mmse_mean, "Low dose", "Alzheimer's", "Male", 17, "High Dose", "Alzheimer's", "Male", 25, "Low dose", "Alzheimer's", "Female", 18, "High Dose", "Alzheimer's", "Female", 24 ) sig_df <- sig_df %>% mutate(drug_treatment = factor(drug_treatment, levels = c("Placebo", "Low dose", "High Dose")), sex = factor(sex, levels = c("Male", "Female")), health_status = factor(health_status, levels = c("Healthy", "Alzheimer's"))) sig_df # plot of cognitive function health and drug treatment g1 <- ggplot(data = sum_df, aes(x = drug_treatment, y = mmse_mean, fill = drug_treatment, group = drug_treatment)) + geom_errorbar(aes(ymin = mmse_mean - mmse_se, ymax = mmse_mean + mmse_se), width = 0.5) + geom_bar(color = "black", stat = "identity", width = 0.7) + facet_grid(sex~health_status) + theme_bw() + scale_fill_manual(values = c("white", "grey", "black")) + theme(legend.position = "NULL", legend.title = element_blank(), axis.title = element_text(size = 20), legend.background = element_blank(), panel.grid.major = element_blank(), panel.grid.minor = element_blank(), axis.text = element_text(size = 12)) + geom_text(data = sig_df, label = "*", size = 8) + labs(x = "\nDrug Treatment", y = "Cognitive Function (MMSE)\n", caption = "\nFigure 1. Effect of novel drug treatment AD-x37 on cognitive function in healthy and demented elderly adults. \nn = 100/treatment group (total n = 600), * indicates significance at p < 0.001") g1 # save the graph! ggsave("ad_publication_graph.png", g1, height = 7, width = 8, units = "in") |
File:Ad public.svg |
Opioid prescribing habits in texas
https://juliasilge.com/blog/texas-opioids/.
- It can read multiple sheets (27 sheets) at a time and merge them by rows.
- case_when(): A general vectorised if. This function allows you to vectorise multiple if_else() statements. How to use the R case_when function.
x %>% mutate(group = case_when( PredScore > quantile(PredScore, .5) ~ 'High', PredScore < quantile(PredScore, .5) ~ 'Low', TRUE ~ NA_character_ ))
- fill()
- bind_rows(). Another example.
- full_join(), left_join(), right_join(), inner_join(). See the exercises from Useful dplyr functions (with examples). Suppose df1=50x3, df2=45x3 with 25 overlaps. Then left_join=50x5, right_join=45x5, inner_join=25x5, full_join=70x5.
- gather()
- replace_na()
- str_to_title()
- count(), Count Observations by Group in R
- How to Count Distinct Values in R n_distinct(), Filtering for Unique Values in R- Using the dplyr
-
top_n(). weight parameter. top_n(n=5, wt=x) won't order rows by weight in the output actually. slice_max(order_by = x, n = 5) does it.
set.seed(1) d <- data.frame( x = runif(90), grp = gl(3, 30) ) > d %>% group_by(grp) %>% top_n(5, wt=x) # A tibble: 15 x 2 # Groups: grp [3] x grp <dbl> <fct> 1 0.908 1 2 0.898 1 3 0.945 1 4 0.992 1 5 0.935 1 6 0.827 2 7 0.794 2 8 0.821 2 9 0.789 2 10 0.861 2 11 0.913 3 12 0.875 3 13 0.892 3 14 0.864 3 15 0.961 3 > d %>% group_by(grp) %>% slice_max(order_by = x, n = 5) # A tibble: 15 x 2 # Groups: grp [3] x grp <dbl> <fct> 1 0.992 1 2 0.945 1 3 0.935 1 4 0.908 1 5 0.898 1 6 0.861 2 7 0.827 2 8 0.821 2 9 0.794 2 10 0.789 2 11 0.961 3 12 0.913 3 13 0.892 3 14 0.875 3 15 0.864 3
Useful dplyr functions (with examples)
- https://sw23993.wordpress.com/2017/07/10/useful-dplyr-functions-wexamples/
- My top 10 favorite dplyr tips and tricks
- Rename columns by using the dplyr select function
- Calculate in row context with dplyr
- Rearrange columns quickly with dplyr everything
- Drop unnecessary columns with dplyr
- Use dplyr count or add_count instead of group_by and summarize
- Replace nested ifelse with dplyr case_when function
- Execute calculations across columns conditionally with dplyr
- Filter by calculation of grouped data inside filter function
- Get top and bottom values by each group with dplyr
- Reflow your dplyr code
Supervised machine learning case studies in R
Supervised machine learning case studies in R - A Free, Interactive Course Using Tidy Tools.
Time series data
- Automating update of a fiscal database for the Euro Area
- readxl::read_excel()
- transmute() (transmute() adds new variables and drops any existing ones), as.Date()
- filter(), is.na()
- na.omit(), first()
- filter(), gather(), bind_rows(), arrange()
- group_by(), summarize()
- rdb(), lubridate::year(), magrittr::%<>%, select(), spread(), mutate(), select(), gather()
- filter(), full_join(), transmute(), !is.na()
- bind_rows(), mutate()
- chain() (deprecated!)
- ungroup()
- tibble(), left_join()
- Exploring eu wide data on new car registrations and co2 efficiency (data is available)
Calculating change from baseline
group_by() + mutate() + ungroup(). We can accomplish the task by using split() + lapply() + do.call().
trial_data_chg <- trial_data %>% arrange(USUBJID, AVISITN) %>% group_by(USUBJID) %>% mutate(CHG = AVAL - AVAL[1L]) %>% ungroup() # If the baseline is missing trial_data_chg2 <- trial_data %>% group_by(USUBJID) %>% mutate( CHG = if (any(AVISIT == "Baseline")) AVAL - AVAL[AVISIT == "Baseline"] else NA ) %>% ungroup()
Split data and fitting models to subsets
https://twitter.com/romain_francois/status/1226967548144635907?s=20
library(dplyr) iris %>% group_by(Species) %>% summarise(broom::tidy(lm(Petal.Length ~ Sepal.Length))
Show all possible group combinations
Ten Tremendous Tricks in the Tidyverse
https://youtu.be/NDHSBUN_rVU (video).
- count(),
- add_count(),
- summarize() w/ a list column,
- fct_reorder() + geom_col() + coord_flip(),
- fct_lump(),
- scale_x/y_log10(),
- crossing(),
- separate(),
- extract().
Gapminder dataset
Hands-on R and dplyr – Analyzing the Gapminder Dataset
Install on Ubuntu
sudo apt install r-cran-tidyverse # Ubuntu >= 18.04. However, I get unmet dependencies errors on R 3.5.3. # r-cran-curl : Depends: r-api-3.4 sudo apt-get install r-cran-curl r-cran-openssl r-cran-xml2 # Works fine on Ubuntu 16.04, 18.04, 20.04 sudo apt install libcurl4-openssl-dev libssl-dev libxml2-dev
80 R packages will be installed after tidyverse has been installed.
For RStudio server docker version (Debian 10), I also need to install zlib1g-dev
Install on Raspberry Pi/(ARM based) Chromebook
In additional to the requirements of installing on Ubuntu, I got an error when it is installing a dependent package fs: undefined symbol: pthread_atfork. The fs package version is 1.2.6. The solution is to add one line in fs/src/Makevars file and then install the "fs" package using the source on the local machine.
5 most useful data manipulation functions
- subset() for making subsets of data (natch)
- merge() for combining data sets in a smart and easy way
- melt()-reshape2 package for converting from wide to long data formats. See an example here where we want to combine multiple columns of values into 1 column. melt() is replaced by gather().
- dcast()-reshape2 package for converting from long to wide data formats (or just use tapply()), and for making summary tables
- ddply()-plyr package for doing split-apply-combine operations, which covers a huge swath of the most tricky data operations
Miscellaneous examples using tibble or dplyr packages
Print all columns or rows
- print(x, width = Inf) # all columns
- print(x, n = Inf) # all rows
Move a column to rownames
?tibble::column_to_rownames
# It assumes the input data frame has no row names; otherwise we will get # Error: `df` must be a data frame without row names in `column_to_rownames()` # tibble::column_to_rownames(data.frame(x=letters[1:5], y = rnorm(5)), "x")
Move rownames to a variable
https://tibble.tidyverse.org/reference/rownames.html
tibble::rownames_to_column(trees, "newVar") # Still a data frame
Old way add_rownames()
data.frame(x=1:5, y=2:6) %>% magrittr::set_rownames(letters[1:5]) %>% add_rownames("newvar") # tibble object
Remove rows or columns only containing NAs
Surgically removing specific rows or columns that only contains `NA`s
library(dplyr) df <- tibble(x = c(NA, NA, NA), y = c(2, 3, NA), z = c(NA, 5, NA) ) # removing columns where all elements are NA df %>% select(where(~ !all(is.na(.x)))) # removing rows where all elements are NA df %>% filter(if_any(.fns = ~ !is.na(.x)))
Rename variables
dplyr::rename(os, newName = oldName)
Drop/remove a variable/column
select(df, -x) # 'x' is the name of the variable
Drop a level
group_by() has a .drop argument so you can also group by factor levels that don't appear in the data. See this example.
Remove rownames
tibble has_rownames(), rownames_to_column(), column_to_rownames()
has_rownames(mtcars) #> [1] TRUE # Remove row names remove_rownames(mtcars) %>% has_rownames() #> [1] FALSE
> tibble::has_rownames(trees) [1] FALSE > rownames(trees) [1] "1" "2" "3" "4" "5" "6" "7" "8" "9" "10" "11" "12" "13" "14" "15" [16] "16" "17" "18" "19" "20" "21" "22" "23" "24" "25" "26" "27" "28" "29" "30" [31] "31" > rownames(trees) <- NULL > rownames(trees) [1] "1" "2" "3" "4" "5" "6" "7" "8" "9" "10" "11" "12" "13" "14" "15" [16] "16" "17" "18" "19" "20" "21" "22" "23" "24" "25" "26" "27" "28" "29" "30" [31] "31"
relocate: change column order
# Move Petal.Width column to appear next to Sepal.Width iris %>% relocate(Petal.Width, .after = Sepal.Width) %>% head() # Move Petal.Width to the last column iris %>% relocate(Petal.Width, .after = last_col()) %>% head()
pull: extract a single column
x <- iris %>% filter(Species == 'setosa') %>% select(Sepal.Length) %>% pull() # x <- iris %>% filter(Species == 'setosa') %>% pull(Sepal.Length) # x <- iris %>% filter(Species == 'setosa') %>% .$Sepal.Length y <- iris %>% filter(Species == 'virginica') %>% select(Sepal.Length) %>% pull() t.test(x, y)
select(): extract multiple columns
slice(): select rows by index
mtcars %>% slice_max(mpg, n = 1) # mpg cyl disp hp drat wt qsec vs am gear carb # Toyota Corolla 33.9 4 71.1 65 4.22 1.835 19.9 1 1 4 1 mtcars %>% slice(which.max(mpg)) # mpg cyl disp hp drat wt qsec vs am gear carb # Toyota Corolla 33.9 4 71.1 65 4.22 1.835 19.9 1 1 4 1
reorder
iris %>% ggplot(aes(x=Species, y = Sepal.Width)) + geom_boxplot() + xlab=("Species") # reorder the boxplot based on the Species' median iris %>% ggplot(aes(x=reorder(Species, Sepal.Width, FUN = median), y=Sepal.Width)) + geom_boxplot() + xlab=("Species")
Anonymous functions
- https://dplyr.tidyverse.org/reference/funs.html
- Is the role of `~` tilde in dplyr limited to non-standard evaluation?
- Use of ~ (tilde) in R programming Language
- lapply and anonymous functions
- dplyr across: First look at a new Tidyverse function.
- Apply a function (or functions) across multiple columns. across(), if_any(), if_all().
- Select variables that match a pattern. starts_with(), ends_with(), contains(), matches(), num_range().
- data %>% group_by(Var1) %>% summarise(across(contains("SomeKey"), mean, na.rm = TRUE))
ny <- filter(cases, State == "NY") %>% select(County = `County Name`, starts_with(c("3", "4"))) daily_totals <- ny %>% summarize( across(starts_with("4"), sum) ) median_and_max <- list( med = ~median(.x, na.rm = TRUE), max = ~max(.x, na.rm = TRUE) ) april_median_and_max <- ny %>% summarize( across(starts_with("4"), median_and_max) )
# across(.cols = everything(), .fns = NULL, ..., .names = NULL) # Rounding the columns Sepal.Length and Sepal.Width iris %>% as_tibble() %>% mutate(across(c(Sepal.Length, Sepal.Width), round)) iris %>% summarise(across(contains("Sepal"), ~mean(.x, na.rm = TRUE))) # filter rows iris %>% filter(if_any(ends_with("Width"), ~ . > 4)) iris %>% select(starts_with("Sepal")) iris %>% select(starts_with(c("Petal", "Sepal"))) iris %>% select(contains("Sepal"))
Reading and writing data
Speeding up Reading and Writing in R
data.table
Fast aggregation of large data (e.g. 100GB in RAM or just several GB size file), fast ordered joins, fast add/modify/delete of columns by group using no copies at all, list columns and a fast file reader (fread).
Note: data.table has its own ways (cf base R and dplyr) to subset columns.
Some resources:
- https://www.rdocumentation.org/packages/data.table/versions/1.12.0
- cookbook
- R Packages: dplyr vs data.table
- Comparing Common Operations in dplyr and data.table
- Cheat sheet from RStudio
- Reading large data tables in R. fread(FILENAME)
- Note that 'x[, 2] always return 2. If you want to do the thing you want, use x[, 2, with=FALSE] or x[, V2] where V2 is the header name. See the FAQ #1 in data.table.
- Understanding data.table Rolling Joins
- Intro to The data.table Package
- Subsetting rows and/or columns
- Alternative to using tapply(), aggregate(), table() to summarize data
- Similarities to SQL, DT[i, j, by]
- R : data.table (with 50 examples) from ListenData
- Describe Data
- Selecting or Keeping Columns
- Rename Variables
- Subsetting Rows / Filtering
- Faster Data Manipulation with Indexing
- Performance Comparison
- Sorting Data
- Adding Columns (Calculation on rows)
- How to write Sub Queries (like SQL)
- Summarize or Aggregate Columns
- GROUP BY (Within Group Calculation)
- Remove Duplicates
- Extract values within a group
- SQL's RANK OVER PARTITION
- Cumulative SUM by GROUP
- Lag and Lead
- Between and LIKE Operator
- Merging / Joins
- Convert a data.table to data.frame
- R Tutorial: data.table from dezyre.com
- Syntax: DT[where, select|update|do, by]
- Keys and setkey()
- Fast grouping using j and by: DT[,sum(v),by=x]
- Fast ordered joins: X[Y,roll=TRUE]
- In the Introduction to data.table vignette, the data.table::order() function is SLOWER than base::order() from my Odroid xu4 (running Ubuntu 14.04.4 trusty on uSD)
odt = data.table(col=sample(1e7)) (t1 <- system.time(ans1 <- odt[base::order(col)])) ## uses order from base R # user system elapsed # 2.730 0.210 2.947 (t2 <- system.time(ans2 <- odt[order(col)])) ## uses data.table's order # user system elapsed # 2.830 0.215 3.052 (identical(ans1, ans2)) # [1] TRUE
- Boost Your Data Munging with R
- rbindlist(). One problem, it uses too much memory. In fact, when I try to analyze R package downloads, the command "dat <- rbindlist(logs)" uses up my 64GB memory (OS becomes unresponsive).
- Convenience features of fread
- The ultimate R data.table cheat sheet from infoworld
OpenMP enabled compiler for Mac. This instruction works on my Mac El Capitan (10.11.6) when I need to upgrade the data.table version from 1.11.4 to 1.11.6.
Question: how to make use multicore with data.table package?
dtplyr
https://www.tidyverse.org/blog/2019/11/dtplyr-1-0-0/
reshape & reshape2 (superceded by tidyr package)
- Data Shape Transformation With Reshape()
- Use acast() function in reshape2 package. It will convert data.frame used for analysis to a table-like data.frame good for display.
- http://lamages.blogspot.com/2013/10/creating-matrix-from-long-dataframe.html
tidyr
Missing values
Handling Missing Values in R using tidyr
Pivot
- pivot vignette
pivot_wider()R> d2 <- tibble(o=rep(LETTERS[1:2], each=3), n=rep(letters[1:3], 2), v=1:6); d2 # A tibble: 6 × 3 o n v <chr> <chr> <int> 1 A a 1 2 A b 2 3 A c 3 4 B a 4 5 B b 5 6 B c 6 R> d1 <- d2%>% pivot_wider(names_from=n, values_from=v); d1 # A tibble: 2 × 4 o a b c <chr> <int> <int> <int> 1 A 1 2 3 2 B 4 5 6
R> d1 %>% pivot_longer(!o, names_to = 'n', values_to = 'v') # Pivot all columns except 'o' column # A tibble: 6 × 3 o n v <chr> <chr> <int> 1 A a 1 2 A b 2 3 A c 3 4 B a 4 5 B b 5 6 B c 6
- In addition to the names_from and values_from columns, the data must have other columns
- For each (combination) of unique value from other columns, the values from names_from variable must be unique
- Conversion from gather() to pivot_long()
gather(df, key=KeyName, value = valueName, col1, col2, ...) # No quotes around KeyName and valueName pivot_long(df, cols, name_to = "keyName", value_to = "valueName")
- From gather to pivot. pivot_longer()/pivot_wider()
- Data Pivoting with tidyr
- A Tidy Transcriptomics introduction to RNA-Seq analyses
data %>% pivot_longer(cols = c("counts", "counts_scaled"), names_to = "source", values_to = "abundance")
- Using R: setting a colour scheme in ggplot2. Note the new (default) column names value and name after the function pivot_longer(data, cols).
set(1) dat1 <- data.frame(y=rnorm(10), x1=rnorm(10), x2=rnorm(10)) dat2 <- pivot_longer(dat1, -y) head(dat2, 2) # A tibble: 2 x 3 y name value <dbl> <chr> <dbl> 1 -1.28 x1 0.717 2 -1.28 x2 -0.320 dat3 <- pivot_wider(dat2) range(dat1 - dat3)
- pivot_longer()’s Advantage Over gather()
- How to carry column metadata in pivot_longer
- Working with Really Wide Data
Benchmark
An evolution of reshape2. It's designed specifically for data tidying (not general reshaping or aggregating) and works well with dplyr data pipelines.
- vignette("tidy-data") & Cheat sheet
- Main functions
- Reshape data: gather() & spread(). These two will be deprecated
- Break apart or combine columns/Split cells: separate() & unite()
- Handle missing: drop_na() & fill() & replace_na()
- Other functions
- tidyr::separate() function. If a cell contains many elements separated by ",", we can use this function to create more columns. The opposite function is unite().
- tidyr::separate_rows(). If a cell contains many elements separated by ",", we can use this function to create one more row. See the cheat sheet link above.
- http://blog.rstudio.org/2014/07/22/introducing-tidyr/
- http://rpubs.com/seandavi/GEOMetadbSurvey2014
- http://timelyportfolio.github.io/rCharts_factor_analytics/factors_with_new_R.html
- tidyr vs reshape2
- A tidyr Tutorial from U of Virginia
- Benchmarking cast in R from long data frame to wide matrix
Make wide tables long with gather() (see 6.3.1 of Efficient R Programming)
library(tidyr) library(efficient) data(pew) # wide table dim(pew) # 18 x 10, (religion, '<$10k', '$10--20k', '$20--30k', ..., '>150k') pewt <- gather(data = pew, key = Income, value = Count, -religion) dim(pew) # 162 x 3, (religion, Income, Count) args(gather) # function(data, key, value, ..., na.rm = FALSE, convert = FALSE, factor_key = FALSE)
where the three arguments of gather() requires:
- data: a data frame in which column names will become row values. If the data is a matrix, use %>% as.data.frame() beforehand.
- key: the name of the categorical variable into which the column names in the original datasets are converted.
- value: the name of cell value columns
In this example, the 'religion' column will not be included (-religion).
dplyr, plyr packages
- plyr package suffered from being slow in some cases. dplyr addresses this by porting much of the computation to C++. Another additional feature is the ability to work with data stored directly in an external database. The benefits of doing this are the data can be managed natively in a relational database, queries can be conducted on that database, and only the results of query returned.
- It's amazing the things one can do in base R (without installing or loading any other #rstats packages)
- Essential functions: 3 rows functions, 3 column functions and 1 mixed function.
select, mutate, rename, recode +------------------+ filter + + arrange + + group_by + + drop_na + + ungroup + summarise + +------------------+
- These functions works on data frames and tibble objects. Note stats package also has a filter() function for time series data. If we have not loaded the dplyr package, the filter() function below will give an error (count() also is from dplyr).
iris %>% filter(Species == "setosa") %>% count() head(iris %>% filter(Species == "setosa") %>% arrange(Sepal.Length))
- dplyr tutorial from PH525x series (Biomedical Data Science by Rafael Irizarry and Michael Love). For select() function, some additional options to select columns based on a specific criteria include
- start_with()/ ends_with() = Select columns that start/end with a character string
- contains() = Select columns that contain a character string
- matches() = Select columns that match a regular expression
- one_of() = Select columns names that are from a group of names
- Data Transformation in the book R for Data Science. Five key functions in the dplyr package:
- Filter rows: filter(). filter is faster than subset() for very large records. But subset() can both subset rows and select columns.
- Arrange rows: arrange()
- Select columns: select(). Or use $ or [[Number]] or [[NAME]].
- Add new variables: mutate()
- Grouped summaries: group_by() & summarise()
# filter jan1 <- filter(flights, month == 1, day == 1) filter(flights, month == 11 | month == 12) filter(flights, arr_delay <= 120, dep_delay <= 120) df <- tibble(x = c(1, NA, 3)) filter(df, x > 1) filter(df, is.na(x) | x > 1) # arrange arrange(flights, year, month, day) arrange(flights, desc(arr_delay)) # select select(flights, year, month, day) select(flights, year:day) select(flights, -(year:day)) # mutate flights_sml <- select(flights, year:day, ends_with("delay"), distance, air_time ) mutate(flights_sml, gain = arr_delay - dep_delay, speed = distance / air_time * 60 ) # if you only want to keep the new variables transmute(flights, gain = arr_delay - dep_delay, hours = air_time / 60, gain_per_hour = gain / hours ) # summarise() by_day <- group_by(flights, year, month, day) summarise(by_day, delay = mean(dep_delay, na.rm = TRUE)) # pipe. Note summarise() can return more than 1 variable. delays <- flights %>% group_by(dest) %>% summarise( count = n(), dist = mean(distance, na.rm = TRUE), delay = mean(arr_delay, na.rm = TRUE) ) %>% filter(count > 20, dest != "HNL") flights %>% group_by(year, month, day) %>% summarise(mean = mean(dep_delay, na.rm = TRUE))
- Efficient R Programming
- Data wrangling: Transformation from R-exercises.
- Express Intro to dplyr by rollingyours.
- the dot.
matrix(rnorm(12),4, 3) %>% .[1:2, 1:2]
- stringr and plyr A data.frame is pretty much a list of vectors, so we use plyr to apply over the list and stringr to search and replace in the vectors.
- https://randomjohn.github.io/r-maps-with-census-data/ dplyr and stringr are used
- 5 interesting subtle insights from TED videos data analysis in R
- What is tidy eval and why should I care?
- The Seven Key Things You Need To Know About dplyr 1.0.0
select()
Select columns from a data frame
select(my_data_frame, column_one, column_two, ...) select(my_data_frame, new_column_name = current_column, ...) select(my_data_frame, column_start:column_end) select(my_data_frame, index_one, index_two, ...) select(my_data_frame, index_start:index_end)
select() + everything()
If we want one particular column (say the dependent variable y) to appear first or last in the dataset. We can use the everything().
iris %>% select(Species, everything()) %>% head() iris %>% select(-Species, everything()) %>% head() # put Species to the last col
group_by()
group_by() + summarise(), arrange(desc())
Data transformation from R for Data Science
- group_by(var1) %>% summarise(varY = mean(var2)) %>% ggplot(aes(x = varX, y = varY, fill = varF)) + geom_bar(stat = "identity") + theme_classic()
- summarise(newvar = sum(var1) / sum(var2))
- arrange(desc(var1, var2))
- Distinct number of observation: n_distinct()
- Count the number of rows: n()
- nth observation of the group: nth()
- First observation of the group: first()
- Last observation of the group: last()
group_by() + nest(), mutate(, map()), unnest(), list-columns
Many models from R for Data Science
- ?unnest, vignette('rectangle'), vignette('nest') & vignette('pivot')
tibble(x = 1:2, y = list(1:4, 2:3)) %>% unnest(y) %>% group_by(x) %>% nest() # returns to tibble(x = 1:2, y = list(1:4, 2:3)) with 'groups' information
- annotate boxplot in ggplot2
- Coding in R: Nest and map your way to efficient code
group_by() + nest() mutate(, map()) unnest() data --------------------> ---------------> ------->
install.packages('gapminder'); library(gapminder) gapminder_nest <- gapminder %>% group_by(country) %>% nest() # country, data # each row of 'data' is a tibble gapminder_nest$data[[1]] # tibble 57 x 8 gapminder_nest <- gapminder_nest %>% mutate(pop_mean = map(.x = data, ~mean(.x$pop, na.rm = T))) # country, data, pop_mean gapminder_nest %>% unnest(pop_mean) # country, data, pop_mean gapminder_plot <- gapminder_nest %>% unnest(pop_mean) %>% select(country, pop_mean) %>% ungroup() %>% top_n(pop_mean, n = -10) %>% mutate(pop_mean = pop_mean/10^3) gapminder_plot %>% ggplot(aes(x = reorder(country, pop_mean), y = pop_mean)) + geom_point(colour = "#FF6699", size = 5) + geom_segment(aes(xend = country, yend = 0), colour = "#FF6699") + geom_text(aes(label = round(pop_mean, 0)), hjust = -1) + theme_minimal() + labs(title = "Countries with smallest mean population from 1960 to 2016", subtitle = "(thousands)", x = "", y = "") + theme(legend.position = "none", axis.text.x = element_blank(), plot.title = element_text(size = 14, face = "bold"), panel.grid.major.y = element_blank()) + coord_flip() + scale_y_continuous()
- Tidy analysis from tidymodels
- Is nest() + mutate() + map() + unnest() really the best alternative to dplyr::do()
filter by date
What Is the Best Way to Filter by Date in R?
ave() - Adding a column of means by group to original data
mutate + replace() or ifelse()
- mutate() is similar to base::within()
- Change value of variable with dplyr
mtcars %>% mutate(mpg=replace(mpg, cyl==4, NA)) %>% as.data.frame() # VS mtcars$mpg[mtcars$cyl == 4] <- NA
- using ifelse()
- using case_when()
- Mutate multiple columns
- Apply the mutate function to multiple columns at once | mutate_at / mutate_all / mutate_if
mutate_at(data, .vars = vars(starts_with("Petal")), .funs = ~ . * 2) %>% head() mutate_at(data, .vars = vars(starts_with("Petal")), `*`, 2) %>% head()
- recode()
char_vec <- sample(c("a", "b", "c"), 10, replace = TRUE) recode(char_vec, a = "Apple", b = "Banana", .default = NA_character_)
inner_join, left_join, full_join
- Mutating joins
- Join Data Frames with the R dplyr Package (9 Examples)
- Join in r: how to join (merge) data frames (inner, outer, left, right) in R
- Dplyr Tutorial: Merge and Join Data in R with Examples
plyr::rbind.fill()
Videos
- Hands-on dplyr tutorial for faster data manipulation in R by Data School. At time 17:00, it compares the %>% operator, with() and aggregate() for finding group mean.
- https://youtu.be/aywFompr1F4 (shorter video) by Roger Peng
- https://youtu.be/8SGif63VW6E by Hadley Wickham
- Tidy eval: Programming with dplyr, tidyr, and ggplot2. Bang bang "!!" operator was introduced for use in a function call.
- JULIA SILGE
- “Do More with R” video tutorials by Sharon Machlis
- Learning the R Tidyverse from lynda.com
dbplyr
- https://dbplyr.tidyverse.org/articles/dbplyr.html
- translate_sql() Translate an R expression to sql. Some examples.
stringr
- stringr is part of the tidyverse but is not a core package. You need to load it separately.
- Handling Strings with R(ebook) by Gaston Sanchez.
- https://www.rstudio.com/wp-content/uploads/2016/09/RegExCheatsheet.pdf
- stringr Cheat sheet (2 pages, this will immediately download the pdf file)
- Detect Matches: str_detect(), str_which(), str_count(), str_locate()
- Subset: str_sub(), str_subset(), str_extract(), str_match()
- Manage Lengths: str_length(), str_pad(), str_trunc(), str_trim()
- Mutate Strings: str_sub(), str_replace(), str_replace_all(), str_remove()
- Case Conversion: str_to_lower(), str_to_upper(), str_to_title()
- Joint and Split: str_c(), str_dup(), str_split_fixed(), str_glue(), str_glue_date()
- Efficient data carpentry → Regular expressions from Efficient R programming book by Gillespie & Lovelace.
str_replace()
Replace a string in a column: dplyr::across() & str_replace()
df <- data.frame(country=c('India', 'USA', 'CHINA', 'Algeria'), position=c('1', '1', '2', '3'), points=c(22, 25, 29, 13)) df %>% mutate(across('country', str_replace, 'India', 'Albania')) df %>% mutate(across('country', str_replace, 'A|I', ''))
split
Split Data Frame Variable into Multiple Columns in R (3 Examples)
Three ways:
- base::strsplit(x, CHAR)
- stringr::str_split_fixed(x, CHAR, 2)
- tidyr::separate(x, c("NewVar1", "NewVar2"), CHAR)
x <- c("a-1", "b-2", "c-3") stringr::str_split_fixed(x, "-", 2) # [,1] [,2] # [1,] "a" "1" # [2,] "b" "2" # [3,] "c" "3" tidyr::separate(data.frame(x), x, c('x1', 'x2'), "-") # The first argument must be a data frame # The 2nd argument is the column names # x1 x2 # 1 a 1 # 2 b 2 # 3 c 3
magrittr
- Vignettes
- How does the pipe operator actually work?
- magrittr and wrapr Pipes in R, an Examination. Instead of nested statements, it is using pipe operator %>%. So the code is easier to read. Impressive!
x %>% f # f(x) x %>% f(y) # f(x, y) x %>% f(arg=y) # f(x, arg=y) x %>% f(z, .) # f(z, x) x %>% f(y) %>% g(z) # g(f(x, y), z) x %>% select(which(colSums(!is.na(.))>0)) # remove columns with all missing data x %>% select(which(colSums(!is.na(.))>0)) %>% filter((rowSums(!is.na(.))>0)) # remove all-NA columns _and_ rows
suppressPackageStartupMessages(library("dplyr")) starwars %>% filter(., height > 200) %>% select(., height, mass) %>% head(.) # instead of starwars %>% filter(height > 200) %>% select(height, mass) %>% head
iris$Species iris[["Species"]] iris %>% `[[`("Species") iris %>% `[[`(5) iris %>% subset(select = "Species")
- Split-apply-combine: group + summarize + sort/arrange + top n. The following example is from Efficient R programming.
data(wb_ineq, package = "efficient") wb_ineq %>% filter(grepl("g", Country)) %>% group_by(Year) %>% summarise(gini = mean(gini, na.rm = TRUE)) %>% arrange(desc(gini)) %>% top_n(n = 5)
- Writing Pipe-friendly Functions
- http://rud.is/b/2015/02/04/a-step-to-the-right-in-r-assignments/
- http://rpubs.com/tjmahr/pipelines_2015
- http://danielmarcelino.com/i-loved-this-crosstable/
- http://moderndata.plot.ly/using-the-pipe-operator-in-r-with-plotly/
- RMSE
f <- function(x) { (y - x) %>% '^'(2) %>% sum %>% '/'(length(x)) %>% sqrt %>% round(2) }
# Examples from R for Data Science-Import, Tidy, Transform, Visualize, and Model diamonds <- ggplot2::diamonds diamonds2 <- diamonds %>% dplyr::mutate(price_per_carat = price / carat) pryr::object_size(diamonds) pryr::object_size(diamonds2) pryr::object_size(diamonds, diamonds2) rnorm(100) %>% matrix(ncol = 2) %>% plot() %>% str() rnorm(100) %>% matrix(ncol = 2) %T>% plot() %>% str() # 'tee' pipe # %T>% works like %>% except that it returns the lefthand side (rnorm(100) %>% matrix(ncol = 2)) # instead of the righthand side. # If a function does not have a data frame based api, you can use %$%. # It explodes out the variables in a data frame. mtcars %$% cor(disp, mpg) # For assignment, magrittr provides the %<>% operator mtcars <- mtcars %>% transform(cyl = cyl * 2) # can be simplified by mtcars %<>% transform(cyl = cyl * 2)
- The Four Pipes of magrittr and lambda functions.
Upsides of using magrittr: no need to create intermediate objects, code is easy to read.
When not to use the pipe
- your pipes are longer than (say) 10 steps
- you have multiple inputs or outputs
- Functions that use the current environment: assign(), get(), load()
- Functions that use lazy evaluation: tryCatch(), try()
Dollar sign .$
- A Short Tutorial about Magrittr’s Pipe Operator and Placeholders, Simplify Your Code with %>%
gapminder %>% dplyr::filter(continent == "Asia") %>% {stats::cor(.$lifeExp, .$gdpPercap)} gapminder %>% dplyr::filter(continent == "Asia") %$% {stats::cor(lifeExp, gdpPercap)} gapminder %>% dplyr::mutate(continent = ifelse(.$continent == "Americas", "Western Hemisphere", .$continent))
- Another example Introduction to the msigdbr package
m_list = m_df %>% split(x = .$gene_symbol, f = .$gs_name) m_list2 = m_df %$% split(x = gene_symbol, f = gs_name) all.equal(m_list, m_list2) # [1] TRUE
- Use $ dollar sign at end of of an R magrittr pipeline to return a vector
DF %>% filter(y > 0) %>% .$y
%$%
Expose the names in lhs to the rhs expression. This is useful when functions do not have a built-in data argument.
lhs %$% rhs # lhs: A list, environment, or a data.frame. # rhs: An expression where the names in lhs is available. iris %>% subset(Sepal.Length > mean(Sepal.Length)) %$% cor(Sepal.Length, Sepal.Width)
set_rownames() and set_colnames()
https://stackoverflow.com/a/56613460, https://www.rdocumentation.org/packages/magrittr/versions/1.5/topics/extract
data.frame(x=1:5, y=2:6) %>% magrittr::set_rownames(letters[1:5]) cbind(1:5, 2:6) %>% magrittr::set_colnames(letters[1:2])
purrr: : Functional Programming Tools
While there is nothing fundamentally wrong with the base R apply functions, the syntax is somewhat inconsistent across the different apply functions, and the expected type of the object they return is often ambiguous (at least it is for sapply…). See Learn to purrr.
- https://purrr.tidyverse.org/
- cheatsheet
- purrr cookbook
- Higher-order function
- Python Decorator/metaprogramming
- Iterating over the lines of a data.frame with purrr
- Functional programming (cf Object-Oriented Programming)
- What does the tilde mean in this context of R code, What is meaning of first tilde in purrr::map
- Getting started with the purrr package in R, especially the map() and map_df() functions.
library(purrr) # map() is a replacement of lapply() # lapply(dat, function(x) mean(x$Open)) map(dat, function(x)mean(x$Open)) # map allows us to bypass the function function. # Using a tilda (~) in place of function and a dot (.) in place of x map(dat, ~mean(.$Open)) # map allows you to specify the structure of your output. map_dbl(dat, ~mean(.$Open)) # map2() is a replacement of mapply() # mapply(function(x,y)plot(x$Close, type = "l", main = y), x = dat, y = stocks) map2(dat, stocks, ~plot(.x$Close, type="l", main = .y))
- map_dfr() function from "The Joy of Functional Programming (for Data Science)" with Hadley Wickham. It can be used to replace a loop.
data <- map(paths, read.csv) data <- map_dfr(paths, read.csv, id = "path") out1 <- mtcars %>% map_dbl(mean, na.rm = TRUE) out2 <- mtcars %>% map_dbl(median, na.rm = TRUE)
- Purr yourself into a math genius
- Write & Read Multiple Excel files with purrr
- Handling errors using purrr's possibly() and safely()
- How to Automate Exploratory Analysis Plots
- Easy error handling in R with purrr’s possibly
- Learn to purrr. Lots of good information like tilde-dot is a shorthand for functions.
function(x) { x + 10 } # is the same as ~{.x + 10} map_dbl(c(1, 4, 7), ~{.x + 10})
- A closer look at replicate() and purrr::map() for simulations
twogroup_fun = function(nrep = 10, b0 = 5, b1 = -2, sigma = 2) { ngroup = 2 group = rep( c("group1", "group2"), each = nrep) eps = rnorm(ngroup*nrep, 0, sigma) growth = b0 + b1*(group == "group2") + eps growthfit = lm(growth ~ group) growthfit } sim_lm = replicate(5, twogroup_fun(), simplify = FALSE ) str(sim_lm, max.level = 1) map_dbl(sim_lm, ~summary(.x)$r.squared) # Same as function(x) {} style map_dbl(sim_lm, function(x) summary(x)$r.squared) # Same as sapply() sapply(sim_lm, function(x) summary(x)$r.squared) map_dfr(sim_lm, broom::tidy, .id = "model")
- Functional programming from Advanced R.
- Functional Programming : Sara Altman, Bill Behrman, Hadley Wickham
- Some learnings from functional programming you can use to write safer programs
.x symbol
Function argument naming conventions (`.x` vs `x`). Se purrr::map
negate()
How to select non-numeric columns using dplyr::select_if
library(tidyverse) iris %>% select_if(negate(is.numeric))
forcats
https://forcats.tidyverse.org/
JAMA retraction after miscoding – new Finalfit function to check recoding
outer()
Genomic sequence
- chartr
> yourSeq <- "AAAACCCGGGTTTNNN" > chartr("ACGT", "TGCA", yourSeq) [1] "TTTTGGGCCCAAANNN"
broom
broom: Convert Statistical Analysis Objects into Tidy Tibbles
Especially the tidy() function.
R> str(survfit(Surv(time, status) ~ x, data = aml)) List of 17 $ n : int [1:2] 11 12 $ time : num [1:20] 9 13 18 23 28 31 34 45 48 161 ... $ n.risk : num [1:20] 11 10 8 7 6 5 4 3 2 1 ... $ n.event : num [1:20] 1 1 1 1 0 1 1 0 1 0 ... ... R> tidy(survfit(Surv(time, status) ~ x, data = aml)) # A tibble: 20 x 9 time n.risk n.event n.censor estimate std.error conf.high conf.low strata <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <chr> 1 9 11 1 0 0.909 0.0953 1 0.754 x=Maintained 2 13 10 1 1 0.818 0.142 1 0.619 x=Maintained ... 18 33 3 1 0 0.194 0.627 0.664 0.0569 x=Nonmaintained 19 43 2 1 0 0.0972 0.945 0.620 0.0153 x=Nonmaintained 20 45 1 1 0 0 Inf NA NA x=Nonmaintained
lobstr package - dig into the internal representation and structure of R objects
Other packages
Great R packages for data import, wrangling, and visualization
Great R packages for data import, wrangling, and visualization
tidytext
https://juliasilge.shinyapps.io/learntidytext/
tidytuesdayR
- https://github.com/rfordatascience/tidytuesday
- https://cran.r-project.org/web/packages/tidytuesdayR/index.html, Github
install.packages("tidytuesdayR") library("tidytuesdayR") tt_datasets(2020) # get the exact day of the data we want to load coffee_ratings <- tt_load("2020-07-07") print(coffee_ratings) # readme(coffee_ratings)