Twitter Sentiment Analysis for crypto currency
The topic modeling iterations on our cryptocurrency tweet dataset helped us in understanding that with every topic modeling iteration the irrelevant tweets get removed and only relevant tweets are filtered, moreover this practice can be used as a standard data cleaning process when dealing with any tweet dataset which can help in narrowing down the tweets to the topic or problem under analysis. For our analysis, we were successfully able to reduce the tweet dataset of 50,711 to 12,060 tweets which we used for generating sentiments around cryptocurrency in the major North American cities and found out that even after a huge turmoil in the crypto market which is down to around 80% from it's all-time high last year, one would expect that the general sentiment would be very negative but the numbers helped us in uncovering that people still believe in the cryptocurrency and the positive opinion is more as compared to the negative opinion.