class: center, middle, inverse, title-slide # Using the <code>%>%</code> operator ## Teaching with the tidyverse --- class: middle, center, inverse  ## What is the `%>%`? *image courtesy [@LegoRLady](https://twitter.com/LEGO_RLady/status/986661916855754752)* --- class: center, middle, inverse # `%>%` ## Combine multiple operations with the pipe https://r4ds.had.co.nz/transform.html#combining-multiple-operations-with-the-pipe --- class: center, middle, inverse # `%>%` ## The pipe *"dataframe first, dataframe once"* -- ```r library(tidyverse) ``` -- RStudio Keyboard Shortcuts: OSX: `CMD` + `SHIFT` + `M` Else: `CTRL` + `SHIFT` + `M` --- class: middle .pull-left[ .shout[Nesting] a dataframe inside a function is .shout[hard to read]. ```r slice(babynames, 1) ``` ``` # A tibble: 1 x 5 year sex name n prop <dbl> <chr> <chr> <int> <dbl> 1 1880 F Mary 7065 0.0724 ``` <hr> Here, the "sentence" starts with a .shout[verb]. ] -- .pull-right[ .whisper[Piping] a dataframe into a function lets you read .whisper[L to R, top-to-bottom]. ```r babynames %>% slice(1) ``` ``` # A tibble: 1 x 5 year sex name n prop <dbl> <chr> <chr> <int> <dbl> 1 1880 F Mary 7065 0.0724 ``` <hr> Now, the "sentence" starts with a .whisper[noun]. ] --- class: middle .pull-left[ .shout[Sequences of functions] make you read .shout[inside out] ```r slice(filter(babynames, sex == "M"), 1) ``` ``` # A tibble: 1 x 5 year sex name n prop <dbl> <chr> <chr> <int> <dbl> 1 1880 M John 9655 0.0815 ``` ] -- .pull-right[ .whisper[Chaining functions] together lets you read .whisper[L to R, top-to-bottom] ```r babynames %>% filter(sex == "M") %>% slice(1) ``` ``` # A tibble: 1 x 5 year sex name n prop <dbl> <chr> <chr> <int> <dbl> 1 1880 M John 9655 0.0815 ``` ] --- ## Think .whisper["and then"] You can build up a series of pipes ```r *babynames %>% # dataframe first and then... filter(sex == "M") %>% arrange(desc(year)) %>% slice(1) ``` --- ## Think .whisper["and then"] You can build up a series of pipes ```r *babynames %>% # dataframe first and then... * filter(sex == "M") %>% # filter only males and then... arrange(desc(year)) %>% slice(1) ``` --- ## Think .whisper["and then"] You can build up a series of pipes ```r *babynames %>% # dataframe first and then... * filter(sex == "M") %>% # filter only males and then... * arrange(desc(year)) %>% # arrange by year and then... slice(1) ``` --- ## Think .whisper["and then"] You can build up a series of pipes ```r *babynames %>% # dataframe first and then... * filter(sex == "M") %>% # filter only males and then... * arrange(desc(year)) %>% # arrange by year and then... * slice(1) # slice the first row! ``` ``` # A tibble: 1 x 5 year sex name n prop <dbl> <chr> <chr> <int> <dbl> 1 2015 M Noah 19511 0.00963 ``` --- ## Use assignment Using the `<-` assigns this new object a name. We ony have to do this _once_ with the `%>%`! -- .pull-left[ .shout[Bad] ```r boys <- filter(babynames, sex == "M") boys_desc <- arrange(boys, desc(year)) boy_latest <- slice(boys_desc, 1) boy_latest ``` ``` # A tibble: 1 x 5 year sex name n prop <dbl> <chr> <chr> <int> <dbl> 1 2015 M Noah 19511 0.00963 ``` ] -- .pull-right[ .whisper[Good] ```r *boy_latest <- babynames %>% filter(sex == "M") %>% arrange(desc(year)) %>% slice(1) *boy_latest ``` ``` # A tibble: 1 x 5 year sex name n prop <dbl> <chr> <chr> <int> <dbl> 1 2015 M Noah 19511 0.00963 ``` ] --- class: inverse, middle, center <img src="https://www.rstudio.com/wp-content/uploads/2014/04/magrittr.png" width="50%" style="display: block; margin: auto;" /> ## "dataframe first, dataframe once" --- class: middle ## What you don't see ```r babynames %>% * filter(sex == "M") %>% * arrange(desc(year)) %>% * slice(1) ``` ``` # A tibble: 1 x 5 year sex name n prop <dbl> <chr> <chr> <int> <dbl> 1 2015 M Noah 19511 0.00963 ``` <hr> -- .pull-left[ This does the same thing: ```r babynames %>% * filter(.data = ., sex == "M") %>% * arrange(.data = ., desc(year)) %>% * slice(.data = ., 1) ``` ``` # A tibble: 1 x 5 year sex name n prop <dbl> <chr> <chr> <int> <dbl> 1 2015 M Noah 19511 0.00963 ``` ] -- .pull-right[ So does this: ```r babynames %>% * filter(., sex == "M") %>% * arrange(., desc(year)) %>% * slice(., 1) ``` ``` # A tibble: 1 x 5 year sex name n prop <dbl> <chr> <chr> <int> <dbl> 1 2015 M Noah 19511 0.00963 ``` ] --- class: inverse, middle, center  ## Why do I need to know about the dot (`.`)? --- ## Why care about `.`? For many non-`tidyverse` functions (and some `tidyverse` ones too), the dataframe is not _actually_ the first argument 😿. -- .pull-left[ .shout[Bad] ```r # none of these will work t.test(babynames, prop ~ sex) babynames %>% t.test(prop ~ sex) babynames %>% filter(year >= 2000) %>% t.test(prop ~ sex) ``` ] -- .pull-right[ .whisper[Good] ```r # these will all work t.test(prop ~ sex, data = babynames) babynames %>% t.test(prop ~ sex, data = .) babynames %>% filter(year >= 2000) %>% t.test(prop ~ sex, data = .) ``` ] --- class:middle, inverse, center  ## Using the `%>%` in tidyverse style --- From [The tidyverse style guide](https://style.tidyverse.org/pipes.html), the `%>%` should: -- - Have a space before, and a new line after. -- - Be indented by two spaces after the first line where you name the dataframe. -- - Include the `()` after each function name (even if you don't use arguments!) -- .pull-left[ .shout[Bad] ```r babynames %>% filter(sex == "M") %>% slice(1) # with assignment boy_latest <- babynames %>% filter(sex == "M") %>% arrange(desc(year)) %>% slice(1) ``` ] -- .pull-right[ .whisper[Good] ```r babynames %>% filter(sex == "M") %>% arrange(desc(year)) %>% slice(1) # with assignment boy_latest <- babynames %>% filter(sex == "M") %>% arrange(desc(year)) %>% slice(1) ``` ]