Financial Futures Prediction Using Fuzzy Rough Set and Synthetic Minority Oversampling Technique

Table 2

The data preprocessing results of the SMOTE-based approach. Note that the class imbalance .

Multiclassification
Class	Majority class	Minority class	Minority class	Minority class	Minority class
The unbalanced samples before processing
Training period 1	717	99	18	92	21
Class imbalance ratio	\	7.24	39.83	7.79	34.14
Training period 2	770	93	12	80	17
Class imbalance ratio	\	8.28	64.17	9.63	45.29
Training period 3	756	104	11	87	18
Class imbalance ratio	\	7.27	68.73	8.69	42.00
Training period 4	772	90	8	87	11
Class imbalance ratio	\	8.58	96.50	8.87	70.18
The sample numbers after processing by SMOTE
Training period 1	396	297	54	276	63
Class imbalance ratio	\	1.33	7.33	1.43	6.29
Training period 2	372	279	36	240	51
Class imbalance ratio	\	1.33	10.33	1.55	7.29
Training period 3	416	312	33	261	54
Class imbalance ratio	\	1.33	12.61	1.59	7.70
Training period 4	360	270	24	261	33
Class imbalance ratio	\	1.33	15.00	1.38	10.91