Inspiration
A post on Predict.it, betting on how many future tweets a selected Twitter user may post in the next week.
What it does
It predicts the number of tweets that someone will have in the next week based on their previous tweet data.
How I built it
The app loads past tweet information from Twitter (using the Twitter API), which it then feeds into a machine learning algorithm that determines the number of tweets the person will likely have next week. This data then gets displayed on a Heroku Django web app.
Challenges I ran into
Heroku had some problems with a couple of import statements not being found/located within the correct directory. The algorithm for guessing the correct amount of future tweets proved to be difficult and required learning a little more about linear regression models.
Accomplishments that I'm proud of
We made the website from scratch with no preset templates or CSS add-ons (Bootstrap), constructed .py files for the guesser algorithm, and experienced both back-end (Python/Django) and front-end (HTML/CSS) development.
What I learned
Machine Learning had to be learned in order to build this project. Specific functions/features in Python IDE (PyCharm), HTML/CSS, and Django. Learned more about the capabilities of Heroku.
What's next for Cool Beans Twitter Predictor
We plan to implement the ability to allow any Twitter user to be searched up and have their data be analyzed/predicted.
Log in or sign up for Devpost to join the conversation.