Innowise Group created the solution for a cryptocurrency exchange to make more exact forecasts of the rise or fall of the bitcoin. As its rate movement depends highly on the mood of investors, it is necessary to take into account and rationalize some emotional factors as well. In addition to standard methods of predicting exchange rates, the system analyzes a huge amount of data from social networks and media, correlates them with trading data, and gives a better forecast.
What Was Done
Our team of Python Developers and Data Scientists used the NLP algorithm that enabled the system to collect and analyze data from Twitter posts that mention the word "bitcoin." Semantic technology processes the logical structure of sentences to identify the most significant elements in the text and understand the discussions' essence. In the next step, the system collects and stores data and then uses analysis to predict changes in the bitcoin exchange rate for a given period (a week or a month).
The advantage of the system is that it is constantly learning. For instance, it can already recognize tiny details such as personal humor or sarcasm, thus making forecasts even more accurate.
Technologies and tools: Python, spaCy, Gensim, Keras, Scikit-learn.