AI Stock Predictor

Inspiration

-We wanted to challenge our knowledge of Python, AI and databases -We wanted to find a way to determine how successful people on twitter are at predicting stock price changes

What it does

-Scrapes Twitter for tweets predicting the movement of specific stocks -The raw data of each tweet and stock movement is sorted through and uploaded to a database -An AI model receives some of the filtered data and learns how to predict whether a tweet's prediction will be correct -The AI test results are visualized on a graph

How we built it

-All code in the project was done in Python -Twitter API (tweepy and Yahoo Finance) to obtain the raw data -PostGreSQL to create the database and filter the data we obtained -Python Libraries (numpy, sci-kit learn) to create the AI classifier and then train and test the AI -We used a Naive Bayes AI model due to its common use in classification and data filtering. Also, it works well with large data sets -MatPlotLib to graph the results from the AI's predictions regarding what percentage of tweets have a successful prediction

Challenges we ran into

• Yahoo Finance API broke after 7 PM because the markets close
• This caused an issue where the data stored for every stock resets
• Our program could not function with empty data
• The database gave issues when importing the data

Accomplishments that we're proud of

• Using 2 APIs that we had no prior knowledge of
• Developing a database from scratch, with little knowledge
• Developing an AI
• Developing a graph in Python

What we learned

• Teamwork is essential to get a big project done
• Communication is also key to succeeding

What's next for Twitter_Hackathon

• Try completing the sponsored challenges
• Try a different stand alone project

Built With

• ai
• python
• sql
• twitter
• yahoo-finance-api