Thread by @asmae_toumi

I'm gonna start a thread on what I hope will be helpful R tips to wrangle this huge NFL Big Data Bowl data. If you're an advanced R programmer, this is probably not for you but feel free to correct me if I made a mistake or offer better alternatives

#1

slice_sample() if you want to quickly preview what your result might look like using a random sampling of rows in your data

#2

janitor::clean_names() if variable names with random capitalization, spaces and other undesired characters make you sick

with the defaults you can turn gameTimeEastern (😒) into game_time_eastern (😙👌)

#3

lubridate::mdy() to convert a variable into a Date

data %>% mutate(game_date = mdy(game_date))

#4

lubridate::parse_date_time() for inconsistent date formats

players %>%
mutate(birth_date = lubridate::parse_date_time(birth_date,
orders = c("y-m-d", "m/d/y"))

#5

tidyverse::case_when()

#6

janitor::get_dupes()

(watch out with your joins, there are 5 players with the same name)

#7

if you're going to bind all 17 weeks of data into one dataset, save it to disk as a parquet file via {arrow}. from my very unscientific testing with different file formats (rda, fst, feather, rds, tsv.gz), parquet was the fastest read

More on {arrow}: arrow.apache.org/docs/r/

#8

tidyverse::separate()

#9

Another use of tidyverse::separate()