SK-MOEFS Multi-Objective Evolutionary Fuzzy System Library effectiveness as User-Friendly Cryptocurrency Prediction Tool

Abstract

The emergence of Cryptocurrency has long foreshadowed a more accessible exchange market. Cryptocurrency is easy to use and trade, and this ease of access into the market brought newcomers into the Crypto-trading scene. This surge of inexperienced newcomers causes market instability and major loss amongst themselves. AI models, algorithms, and systems have been long used as an important aspect of prediction. However, the use of AI systems is complex. AI tools and systems often use complicated mathematical formulas and are not easily understood. Amongst these AI systems, Fuzzy Rule-Based Systems (FRBSs) has one of the most easily understood displays. With accuracy that rivals of other less-understood methods, such as Neural Network, FRBSs present us a choice that is easily used by users while keeping the interface as basic and simple as possible. This paper aims to study the use of FRBSs using SK-MOEFS (SciKit-Multi Objective Evolutionary Fuzzy System) Python Library in predicting a bull signal or a bear signal in the Cryptocurrency market while still preserving FRBSs user-friendly nature. The fuzzy sets are partitioned as Very Low, Low, Medium, High, and Very High. Then the resulting classification are used to signal whether a Cryptocurrency is bearish or bullish on the current day. The parameter used on the system yields an undesirable result of 53% accuracy with 25 Total Rule Length, however still producing the desired ease-of-use nature of FRBSs.