Inspiration
We realized that we were getting scammed on Amazon for certain products that we wanted because when we looked at the changes in price over a month or 3 months, we would see that the prices fell by as much as $5 or $6 from the average. We wanted to figure out if there is a way to predict price movements based on the price history of the product in order to save money.
What it does
Price Predictor takes in an Amazon link, finds a price chart associated with the product corresponding to that link, and uses LSTM, a recurrent neural network, on the price chart data to predict the price of the product a week later.
How we built it
We used Heroku and Flask to host our model and website, and Python to create the neural network.
Challenges we ran into
Getting Flask and Heroku to work together, and getting the model onto the Heroku platform
Accomplishments that we're proud of
Getting a functional website that can perform the train and prediction operations on price history data
What we learned
How to use Flask and Heroku
What's next for Price Predictor
Expanding it to more products (for now, we tried only a few products that had a varying price history), exploring different neural networks to better increase accuracy, sending an email/text reminder to the user to tell them to check Amazon for an updated price
Slides: https://docs.google.com/presentation/d/1QCZiKojRKLx_8obI245_2oMUgVteyXjRHdKazbRPIYk/edit?usp=sharing
Log in or sign up for Devpost to join the conversation.