A comprehensive evaluation of ensemble learning for stock-market prediction

Journal of Big Data

Table 1 Comparison of related studies

Articles	Base learner	Number of weak learners	Ensemble algorithms (combination method)	Datasets		Machine learning task		Evaluation methods
Articles	Base learner	Number of weak learners	Ensemble algorithms (combination method)	Source	Type	CLF	REG	Evaluation methods
[19]	MLP		BAG	Tokyo stock exchange	Stock	√
[18]	SVM and NN	Not stated	STK	S&P 500 index	Stock	√		Cross-validation
[20]	RNN	Not stated	STK	American Association of Individual Investors	Stock	√		Accuracy
[12]	LR, NN, K-NN and SVM		BAG, BOT	Amadeus database	Stock	√		Operating characteristic curve (AUC)
[23]	SVM	10	MV	Sao Paulo Stock Exchange Index	Stock	√		10-fold CV
[21]	DT, RF and ANN	Not stated	BAG, BOT	U.S stock (S&P 500)	Stock		√	MAPE, RMSE, R²
[30]	RF	200	BAG	NASDAQ	Stock		√	CV, MAPE, RMSE
[22]	RF	30	BAG	NASDAQ		√		Accuracy, precision Recall and specificity
[24]	BMA WALS and LASSO		BOT-BAG	Not stated	stock	√		Out-of-sample R²
[28]	NN	1–5	BAG	Mexican stock exchange	Stock	√
[37]	SVM and extra tress	1–250	STK	Not stated	Stock		√	RMSE
[25]	Tress	Not stated	BOT	Not stated	Stock		√	RMSE, MAPE and MSE
[26]	LSTM	10–50	BOT	S&P 500	Stock		√	MAPE
[41]	SVM, RF	Not stated	Voting	BSE SENSEX	Stock	√		Accuracy
[42]	NN	2–5	Not stated	CIMB stock market	Stock	√		MSE, Accuracy
[44]	SVM, LSTM and Multiple Regression	Not stated	Not stated	Yahoo stock data	Stock	√		Accuracy
[45]	NN	30	Not stated	Brazilian stock market	Stock	√		Precision and recall
[29]	NN	Not stated	BAG	Chinese stock market	Stock	√		Accuracy
[27]	Tress and LSTM	50–150	BOT, STK	S&P500 and Nasdaq	Stock	√		F-score, AUC and accuracy
[4]	RNN	Not stated	STK		Stock	√		AUC, accuracy

CV cross-validation, RNN recurrent neural networks, CLF classification, MV majority voting, REG regression, MAPE mean absolute percentage error, MLP multi-layer perceptron, RMSE root mean square error