Table 4 Relationship between the number of principal components and training efficiency
From: Short-term stock market price trend prediction using a comprehensive deep learning system
Number of features | Training dataset preparation time (s) | Test dataset preparation time (s) | Training time (s) | Sum (s) |
|---|---|---|---|---|
29 selected features | 187.46 | 16.30 | 648.53 | 852.29 |
20 principal components | 160.29 | 14.24 | 602.68 | 777.21 |
15 principal components | 125.20 | 12.18 | 591.93 | 729.31 |
10 principal components | 96.54 | 10.37 | 590.76 | 697.67 |
5 principal components | 59.37 | 8.22 | 572.88 | 640.47 |