Inspiration Imagine, you have the greatest idea for a grocery product and you just don't know how to price it. All of the other products on the market seem to know how to do this effortlessly, so why not put their work to good use?
What it does This model has ingested the titles and corresponding prices for over 5000+ products and can predict the price of an item given only its name. It has achieved an accuracy of 89.745%
How we built it The model is an Adaptive Boost Regressor, that was tasked with finding the market price of an item given only its description/title. It ingested the NCR Catalog data (which mainly has items in the grocery category). The text values were converted to ints and then fed into the model. It then returns a price in USD. The site was developed with Flask as the web server, Marko, and lasso, bootstrap for site design and a healthy dollop of CSS
Challenges we ran into Getting the proper array size for inputting the models. Padding was required, the next challenge was marrying the ML to a site
Accomplishments that we're proud of The model achieved a fairly high efficiency on data that you'd imagine to have only a minor correlation.
What's next for Neural Price Prediction Expanding the prediction service to also predict based on category, and product type.
Log in or sign up for Devpost to join the conversation.