Friday week ago, November 27th, saw this year’s edition of the Late Late Toy Show being aired. An average of 1.36 million people are estimated to have watched the show, rising to around 2 million or 73% of the total audience share. I wasn’t one of these 2 million but I caught some of the replay on Sunday and was kept up to date through the Twitter hashtag latelatetoyshow. Last week I pulled down all tweets (excluding retweets) which used the hashtags latelatetoyshow and toyshow to see if I could find the main topics and tweeters on the night.
Between them almost 45,000 tweets were sent using these tags and these were retweeted a further 55,000 times for a total of 100,000 tweets from Friday morning until Sunday night. Of this figure about 91,000 (49,000 retweets) contained the more popular tag, #latelatetoyshow. I have included #toyshow as well as this was the tag used by Waterford Whispers. Of the 45,000 original tweets, these were sent by a total of 12,458 users.
I followed much the same approach here as I did in an earlier project where I clustered tweets on the Vincent Browne Show. I ran into some of the same problems as I did the last time, ending up with one very large cluster containing over 50% of the tweets when I came across a tutorial on using Latent Dirichlet Allocation by Dublin based Aylien. Using LDA on the corpus of tweets I identified 10 topics:
Topic 1 and 2: Topics containing references to "watching", "Christmas", "time" and "tonight". Tweets talking about "It’s almost time...", "Watching the toyshow tonight", "Time for the books" etc
Topic 3: FearTheDeere and "John" and "Joe", possibly references to horologist John‐Joe.
Topic 4: Bony Pony/Cute/Closing Song ‐ References to the girl who sang "My little Pony", words such as cute, omg, amazing and beast, a reference to the closing song. Also contains oh and god, not always used positively ("oh god, not more singing" etc.)
Topic 5: Books and "We’re trending worldwide"!
Topic 6: Competitions and the audience getting a Millennium Falcon.
Topic 7: Evelyn Cusack, Weather, Johnny, David Walliams, Ed Sheeran and Guest. Users comparing this year’s big guest to last years.
Topics 8 and 9: Ryan Tubridy topics. The first related mainly to the tag #toyshowbingo and Star Wars while the second related to jumpers, ad breaks, coke, cake and stretch marks. The first seems to be "Ryan has done something, mark it off #toyshowbingo" while the second is him reacting to certain things (destroying a cake, a doll with stretch marks) and suspicions about what he was consuming during the commercial breaks.
Topic 10: Rapping Farmer/#WhyIsThereALionInMyField
By repeatedly running LDA with different numbers of topics and max_features, or even the same, different topics appeared and sometimes topics split up differently. Sometimes trad and music would appear as would drone. One of the two Ryan Tubridy groups (there was always at least two) would sometimes contain robot and/or pregnant. Player, referring to the RTE Player, often appeared as well as large numbers, over 100,000, tried to access it from abroad on the night. Evelyn Cusack, Rapping Farmer and Fear The Deere were always represented as topics. I’m not sure why Star Wars and Millennium Falcon ended up separated, sometimes they appear together.
Taking these main topics I plotted a timeline showing every tweet sent (excluding retweets) in blocks of 5 minutes. Large drops coincide with the start, end and commercial breaks. The main words in each of the above topics is also plotted. The topics found do a pretty good job of explaining all the major peaks in the plot.
A lot of major companies used the hashtags as a way of advertising competitions on the night. Excluding them, tweets belonging to RTE accounts and media brands like Joe.ie and Waterford Whispers, these are the top 10 most retweeted tweets by individual users (open links at your own risk!):
Laurasaurus2000: 10 minutes into the Toy Show and Chill #LateLateToyShow https://t.co/9lIlcTYPJj Retweeted: 1017
monoIogues: DID SHE SAY UP THE RA #LateLateToyShow https://t.co/b8TAIwFkm0 Retweeted: 634
KourtneyKenny_: When your da kills nidge so you have to go under witness protection and become a farmer #LateLateToyShow https://t.co/k9lquGQgqK Retweeted: 418
twisteddoodles: My Little Pony Poetry #LateLateToyShow https://t.co/MqRmz3c6Lb Retweeted: 278
RyanCullen90: Great first section to the show. Highlight for me was the little Asian shooting up heroin #LateLateToyShow https://t.co/GvMbQ2jeNL Retweeted: 225
RyanCullen90: "How was the #LateLateToyShow?" "It was cool until Tubridy attacked some kid in a wheelchair" https://t.co/5JeA09fOma Retweeted: 221
davidwalliams: A fantastic day in Dublin. A book signing @easons & then I was the surprise guest on #LateLateToyShow talking about #GrandpasGreatEscape Retweeted: 216
evaprxce: i’ll have what he’s having #LateLateToyShow https://t.co/V5rjt7OpT6 Retweeted: 210
brendan_m96: All set #LateLateToyShow https://t.co/6xkPVNJ88W Retweeted: 206
OwensDamien: They should get Vincent Browne to present the #LateLateToyShow. "What’s this plastic crap?" "It’s a Transf-" "In my day you got an orange." Retweeted: 202
A total of 4223 separate hashtags were used during weekend. Excluding #latelatetoyshow and #toyshow, #bookelves appeared 765 times, #fearthedeere 474 times and #toyshowbingo 470 times. Next was #blackfriday on 278, one of many contained in tweets mentioning the top 5 hashtags trending in Ireland, Britain and Worldwide. #oneforeveryoneintheaudience was mentioned 56 times.
A total of 2171 individual users were mentioned in the body of tweets. David Walliams was mentioned 274 times after his appearance on the show, authors Sarah Webb and Zoella 62 and 61 each and comedian Ryan Cullen got 57 mentions along with Ed Sheeran.
Four users, including Sarah Webb, sent more than 100 tweets in the time period with 151 users sending more than 33 tweets, the same amount I have sent in the last 6 months. By grouping together all tweets made by a user I was able to see which user go the most retweets:
Top 10 most retweeted
User Retweeted Tweets RyanCullen90 1672 39 Laurasaurus2000 1145 2 colmtobin 909 27 monoIogues 635 4 sarahwebbishere 505 115 twisteddoodles 479 5 KourtneyKenny_ 418 1 rickoshea 396 15 DaveyReilly 383 32 GerryMcBride 327 32
I’ll be looking at brand interaction with the Toy Show differently but if you are wondering about Waterford Whispers on the night the account sent a total of 64 tweets, was retweeted 2754 times and got 4542 favourites. Last year they used the #latelatetoyshow but this year went for #toyshow. This tag made up only 10% of the tweets on the night and I’m not sure if this held back their full reach.
Another user on #toyshow was Labour TD Aodhan O’Riordain. His 7 tweets garnered 56 retweets and 190 favourites and at one stage he even declared #EvelynCusackIsGod.
Overall I was happy to have pulled out the main topics of the show and the addition of LDA to my knowledge base is an improvement from the last time I did a project similar to this. I was interested to see the mix of most retweeted users, from users with 1 or 2 thousand followers getting higher numbers then media personalities and users with over 100k followers. If I had pulled down all the retweets as well it would, maybe, be possible to build some form of a network graph of retweeting users or even look at the amount of retweets per 5 minute block. Taking @Laurasaurus2000’s top tweet, the most retweeted on the night other than one relating to a competition by Cadbury, how did the popularity of this tweet grow? Which users were the important ones in it reaching 1000 retweets? Did retweets come in a linear or exponential fashion?
A figure of 12,000 viewers is still less than 1% of the average viewership of the Late Late Toy Show. Many brands ran Twitter competitions on the night and this is possibly something for RTE to look into in the future.