Since the twitter service is unstable and sometimes it does not work, we seperate the data collecting program from the main program. To run the program, we need to first run the python collectTwitterData.py command in terminal in order to generate the dataset for training and testing (train.csv and test.csv). The training data is tweets about Tron sent 3 hours ago from present, and the testing data is tweets sent 1 hour ago from present. Then we can run the python sentiment_analysis.py command. The program will build a sentiment analysis modeling based on the training data, and output the predicted sentiment tendency for the testing data. The sentiment is represented as the number of positive comments/ total number of comments.

Once the program is ran, a twitter bot (@TeamBot123) will tweet the current market direction of Tron (TRX) and the sentiment from the past 1 hour tweets. The current market direction is gotten with a web scraper off of coinmarketcap.com.