[3줄 ML] Comparative Performance of Machine Learning Algorithms for Cryptocurrency Forecasting

1. Summary

• This paper presents a comparative performance of Machine Learning algorithms for cryptocurrency forecasting. Specifically, this paper concentrates on forecasting of time series data => SVM, ANNs, DL, BoostedNN 총 4개의 ML 모델 사용
• The data is prepared from daily open, close, high and low price of a daily trading for all total of six types of cryptocurrencies(BTC, ETH, LTC, NEM, XRP, XLM) and are downloaded from the market capitalization database and range from 2013 through 2018
• SVM has several advantages over the other models in forecasting, and previous research revealed that SVM provides a result that is almost or close to actual result yet also improve the accuracy of the result itself.

2. Research Method

2.1. 학습 및 훈련 데이터 Taxanomy

학습 및 훈련 데이터에 사용할 변수는 시가, 종가, 고가, 저가 총 4개의 변수를 사용.

Variable Description

2.2. 학습 및 훈련 데이터 정의

Training Data, Test Data

2.3. 연구 방법론

연구 방법론의 경우, 일반적인 기계 학습 방법론의 절차(데이터 수집 => 전처리 => 학습 => 예측 및 성능 평가)와 동일하게 진행. 

Proposed Methodology

3. Conclusion

3.1. Prediction Accuracy by ML Models

Performance Measures by various classifiers

• 전체 자산군 및 모델을 조합한 24개의 결과 중에서 ETH에 대한 예측력이 95.5%로 가장 높고, 사용된 모델은 SVM
• 반면, 전체 예측 조합 중에서 가장 낮은 예측력인 47.7% 역시 SVM에 의해 예측된 결과임

3.2. Prediction Accuracy Breakdown by Models

핵심만을 정리하기 위해 시가총액이 가장 높은 BTC에 대한 예측 결과만을 Figure 첨부. 

Bitcoin Prediction Chart

3.3. 학습 데이터의 표본 수 및 전문가 부재

However, recent research has showed that due to small range of samples and data manipulation by inadequate evidence and professional analyzers, overall status and accuracy rate of the forecasting needs to be improved in further studies. Thus, advanced research on the accuracy rate of the forecasted price has to be done

---

Source

• Title : Comparative Performance of Machine Learning Algorithms for Cryptocurrency Forecasting (Indonesian Journal of Electrical Engineering and Computer Science · September 2018)
• Paper: link

(PDF) Comparative Performance of Machine Learning Algorithms for Cryptocurrency Forecasting