Using the Pushshift API for Reddit I create a dataset of 9,000 Onion headlines and 15,000 "Onion-like" headlines from the subreddit r/NotTheOnion.
I then use TensorFlow/Keras to build some different neural networks to try to classify the headlines as real Onion or not and invite others to beat my "benchmark" ;-)
The board is generated using JavaScript and the longest words are found in an efficient manner using a Trie data structure which is explained in more detail beneath the Boggle board.
Here I use reinforcement learning to try to learn the optimal blackjack playing strategy!
In particular, the game is modeled as a Markov Decision Process and then Policy Iteration is performed to learn under what circumstances should someone draw or stop.
The website labsandmore.org uploads dogs in California that are up for adoption but the waitlist fills up extremely fast and there's no notification system or email list.
Thus I've been called to duty to scrape the website every hour and send a text when new dogs have been found for my friend looking to adopt.
In the repo, impossible, I have a notebook where I use Selenium to scrape tweets past what the API allows, and use this scraper to get tweets referencing either Impossible Burgers or Beyond Burgers to run a sentiment analysis to see who Twitter thinks makes the better vegan burger 🥕