Twitter Attribute Classification With Q-Learning on Bitcoin Price Prediction

Bitcoin price prediction based on people’s opinions Twitter usually requires millions of tweets, using different text mining techniques, and developing a machine learning model to perform the prediction. These attempts lead employment significant amount computer power, central processing unit (CPU) utilization, random-access memory (RAM) usage, time. To address this issue, in paper, we consider classification tweet attributes that effects changes resource usage levels while obtaining an accurate classify having high effect movement, collect all Bitcoin-related tweets posted certain period divide them into four categories following attributes: $(i)$ number followers poster, notation="LaTeX">$(ii)$ comments tweet, notation="LaTeX">$(iii)$ likes, notation="LaTeX">$(iv)$ retweets. We separately train test by Q-learning with above categorized sets find best among them. compare our approach classic where are used as input data for model, analyzing CPU workloads, RAM memory, time, accuracy. The results show users most have influence future price, their utilization leads spending 80% less 88.8% consumption, 12.5% more predictions compared approach.