My

Like in 2017 I tweeted too much and therefore was unable to rely onrtweet::get_timeline() (or rtweet::get_my_timeline()) to download mytweets so I exported data via the Tweets tab ofhttps://analytics.twitter.com/. Last year, I downloaded one file perquarter but somehow had to download one per month this time. It wasmonotonous but not horrible.

1 2 3 4 5 6

library("magrittr") dir("data", full.names = TRUE) %>% purrr::map_df(readr::read_csv) %>% unique() -> my_tweets

4196 tweets!

Similarly to my 2017 post, I chose the number of likes as criterion todefine my best tweets.

1 2 3 4 5

my_tweets <- dplyr::arrange(my_tweets, - likes) my_tweets <- janitor::clean_names(my_tweets) knitr::kable(my_tweets[1:9, "likes"])

likes
162
151
128
112
95
94
91
86
75

1 2	best9 <- my_tweets$tweet_permalink[1:9]

The table above shows that my best tweets were not extremely popular.

My 2017 blog post inspired Bob Rudis to contribute a tweet_shot()function to rtweet, relying on the webshot package, so that’s what Iused.

1 2 3 4 5 6 7 8 9 10 11

shot <- function(permalink){ rtweet::tweet_shot(permalink) %>% magick::image_crop(magick::geometry_area(width = 517, height = 517, y_off = 88)) %>% magick::image_border("salmon", "20x20") } images <- purrr::map(as.character(best9), shot)

I improved the collage code with two tweaks:

I created rows then columns instead of the other way round, becausethat’s what Instagram does. I did not pay attention to this lastyear but Andrew Caines told me this in a comment.

I did not save the intermediary images to disk. I’ve upped mymagick game!

1 2 3 4 5 6 7 8 9 10 11 12

make_row <- function(i, images){ images[((i-1)*3+1):((i-1)*3+3)] %>% magick::image_join() %>% magick::image_append() } purrr::map(1:3, make_row, images = images) %>% magick::image_join() %>% magick::image_append(stack = TRUE) %>% magick::image_border("salmon", "20x20")

Worth noting from my R year are, according to these tweets,

What about your R year, on Twitter and elsewhere? Have a good last daysof 2018, and a happy 2019!

Related

---