silentaccordion / tm-twitter-api

Working with the Twitter API

Jordan's Introduction

Tweepy can be used to gather all sorts of information from Twitter. In the notebook, Pulling Data with Tweepy, I have laid out ways to gather follower information as well as Tweets containing a certain key phrase or Twitter handle. One thing to keep in mind while going through the notebook is that Twitter only allows so many pulls per fifteen minute periods, so if you are choosing to grab followers from accounts with hundreds of thousands to millions of followers, it will take hours, and possibly upto days to complete. I showed how to grab certain objects from followers/tweets, but if there's something in particular you think would be interesting (Geo location, favorite/retweet numbers, following counts, etc.) there is almost certainly a way to do it, so feel free to go hunt for it in the Twitter Dev documentation.

If you don't waste hours a day on Twitter like me, here's some things to keep in mind about Twitter (the platform, not the API). Each Tweet can contiain at most 280 characters, so a few NLP-type statistics about text/word length won't lend themselves well to analysis. Each Tweet can also be favorited or retweeted, where retweeting someone elses Tweet will share it to your own followers timeline. Lastly, hashtags are a decent way to follow a topic/conversation. You'll see in the exercise that I grabbed Tweets from the hashtag #WorldMentalHealthDay, but almsot any large organization/event/topic has their own hashtag that they try to encourage others to engage with. Some others include #Debates2020, #NBAFinals, etc.

Twitter Assignment

In this assisgnment we're going to pull a handful of fields for all the followers of several Twitter accounts.

In the code provided by Jordan, he picks two WNBA teams and pulls (some of) the followers screen_names and descriptions. This assignment will extend that work.

Pick at least two Twitter accounts that you'll be comparing. Those accounts should be on different sides of some sort of divide. The divide could be ideological, sports-related, public health-related, age, etc.

Ideally the accounts will have somewhere between 100K and 500K followers. We're picking that range so that the results won't take too long with Twitter's rate limiting. If you'd like to pick a few accounts to represent one "side" that have that many followers in aggregate, that's fine.

Once you've chosen your accounts, write code that pulls all the followers of the accounts (or set of accounts).

• screen_name
• name
• id
• location
• followers_count
• friends_count
• description

These fields are all located in the user object accessed in the code.

Write these fields out to two tab-delimited text files. I recommend following the example code and including the name of the initial account in the file name.

Note that the output of this assignment should be two tab-delimited text files, one for each starting account or set of accounts.

10/22 Update from Jordan

Brenden Connors made some really useful code that grabs all user objects from our list of Twitter IDs. Instead of grabbing 300 records per fifteen minutes, you should now be able to grab about of 900,000.