Deep-learning-assisted business intelligence model for cryptocurrency forecasting using social media sentiment

Abstract

PurposeBusiness Intelligence has gained a significant attraction in the recent past and facilitates managers for efficient business decision-making. Over the years, the attraction toward the cryptocurrency (CC) market has increased. Since the CC market is highly volatile, it is extremely sensitive to shocks and web data related to large events happening around the globe.Design/methodology/approachThis research study provides a business intelligence model to predict five top-performing CCs. In this study, deep learning, linear regression and support vector regression (SVR) are used to predict CC prices. The sentiment of some mega-events is also used to enhance the performance of these models.FindingsThe results show that models of business intelligence such as deep learning and SVR provide better results. Moreover, the results show that the incorporation of social media sentiment data significantly improves the performance of the proposed models. The overall accuracy of the model improves approximately twofold when multiple event sentiments were incorporated.Originality/valueThe use of social media sentiment of global and local events for different countries along with deep learning for CC forecasting.

Business Intelligence (BI) System for Budgeting and Customer Satisfaction

“…Once they implement the system and train their staff members on how to use it, many hours of work can be shortened while maintaining the profit margins. Additionally, it will allow them to know what machines can be replaced to cut cost [17], [18]. With the help of the hardware they can record new data such as age groups, gender and visualise who comes to the casino most often.…”

Effect of twitter investor engagement on cryptocurrencies during the COVID-19 pandemic

“…( Garcia and Schweitzer, 2015 , Kraaijeveld and Smedt, J., 2020 , Mirtaheri et al, 2021 , Öztürk and Bilgiç, 2021 , Wołk, 2020 , Yasir et al, 2022 )…”

Predicting Top Five Cryptocurrency Prices via Linear Structural Time Series (STS) Approach

“…The ARIMA model is considered the most common linear statistical model for time series analysis [33,34]. The models that are most widely used to investigate price fluctuation using econometrics are the generalized autoregressive conditional heteroscedasticity (GARCH) model [27,35], vector autoregressive (VAR) model [36,37], and multiple linear regression [38,39]. Note that these econometric models have specific assumptions needed to be fulfilled.…”