Code Jam: DataDive
By Alexander Torabi, Christos Panaritis, Karine Mellata and Sadhvi Mehta
Data used: bitcoin_dataset.csv, abcnews-date-text (external), nytimes (self collected data) and google news (self collected data).
The idea behind the project was to be able to link bitcoin price changes to news articles. Having scrapped google news and new york times and other news outlets for headlines (as well as there publication date) related and not related to bitcoin. Using the dataset, it was possible to link the rise or decrease in stocks with the important keywords used in headlines. Using TD-IDF, we used a Neural Network to further predict the rise or decrese in a stcok only based on keywords!
This program takes in as input a headline (typed by the user) and returns how that headline might affect the price of bitcoin.
Application: Having an aproximated change in price could be used to help people invest in bitcoin by minimizing losses as well as maximizing profits.