We rate dogs Udacity project
As part of a Udacity Data Analyst Nanodegree I analyzed a Twitter account: We Rate Dogs!.
We rate dogs is a popular Twitter account posting funny pictures and ratings about dogs. I want to analyze the coverage of that account.
The dataset used spans from End of 2015 to Summer 2017, and I learned how to use the twitter api to download the data:
The figure above shows the amount of tweets from that account in the considered time frame and it can be seen, that the were more active in late 2015.
The dog jury of We rate Dogs has their own vocabulary in naming different types of dogs, like ‘doggo’ or ‘pupper’.
In the dataset given, they used it in around a quarter of occasions:
Most of the dogs remained untyped, but around a sixth was classified as a ‘Pupper’ and the least amount of dogs were ‘Floofers’.
Here are two examples for ‘Puppers’, source Twitter, just click the images to link to the tweets:
Meet Benji. He just turned 1. Has already given up on a traditional pupper physique. Just inhaled that thing. 9/10 pic.twitter.com/sLUC4th3Zj
— WeRateDogs® (@dog_rates) June 5, 2016
And here one example for a fluffer:
Coverage
In total the account generated over 10.000.000 retweets and favs, with a higher momentum in the earlier time span:
Another interesting fact within the twitter account of we rate dots, are their (name giving) Dog Rates. They mostly rate dogs within a “… out of ten scale”, so for example 8/10, or 4/10. But sometimes a dog receives only a top note, and on other times the nominator may be higher than 10, so for example 14/10.
Therefore I normalized all values to be within (0, 1) and also added a denominator of 10, wherever it was missing:
So now it is clear, that there are only a few outliers, namely those four guys. If you are a dog person, brace for a cuteness impact:
Maybe as a last annotation it is worth noting, that the cuteness of the dogs increases slightly over time:
so it might be interesting to stay tuned and follow that twitter account, if you are a dog person!