An adaptive feature selection schema using improved technical indicators for predicting stock price movements
Download 2.42 Mb. View original pdf
|
- Navigate this page:
- 4. Methodology
| Table 1 The part of SSEC, HSI, DJI, SP 500 data sets. Data Date Open High Low Close Volume SSEC 2021–03-01 3531.48 3552.57 3511.99 3551.40 31,548,752,600 2021–03-02 3566.85 3566.85 3485.36 3508.60 33,983,048,600 2021–03-03 3500.15 3577.62 3498.72 3576.90 34,765,684,600 2021–03-04 3546.64 3552.20 3487.38 3503.49 39,361,612,000 2021–03-05 3463.31 3523.57 3456.67 3501.99 35,640,922,300 HSI 2021–03-01 29457.89 29550.75 29195.97 29452.57 2,629,062,100 2021–03-02 29708.39 29765.96 28957.31 29095.86 2,895,849,600 2021–03-03 29249.43 29912.00 29183.56 29880.42 3,228,618,000 2021–03-04 29525.48 29597.16 29102.10 29236.79 2,957,909,000 2021–03-05 28667.14 29397.27 28513.13 29098.29 3,996,713,300 DJI 2021–03-01 29457.89 29550.75 29195.97 29452.57 2,629,062,100 2021–03-02 29708.39 29765.96 28957.31 29095.86 2,895,849,600 2021–03-03 29249.43 29912.00 29183.59 29880.42 3,228,618,000 2021–03-04 29525.48 29597.16 29102.10 29236.79 2,957,909,000 2021–03-05 28667.14 29397.27 28513.13 29098.29 3,996,713,300 SP 500 2021–03-01 3842.51 3914.50 3842.51 3901.82 5,071,540,000 2021–03-02 3903.64 3906.41 3868.57 3870.29 5,493,690,000 2021–03-03 3863.99 3874.47 3818.86 3819.72 6,150,790,000 2021–03-04 3818.53 3843.67 3723.34 3768.47 7,142,240,000 2021–03-05 3793.58 3851.69 3730.19 3841.94 6,842,570,000 Table 2 Technical indicators in this study. Indicator Description Indicator Description CP Close Price ROC Rate Of Change MACD Moving Average Convergence/Divergence VAR Variance RSI Relative Strength Index DEMA Double Exponential Moving Average FASTK Stochastic Fast K ATR Average True Range FASTD Stochastic Fast D BETA Beta ULTISC Ultimate Oscillator ADX Average Directional Movement Index PRICE_C Price Change CCI Commodity Channel Index TSF Time Series Forecast OBV On Balance Volume VAR Variance WR Williams R G. Ji et al. Expert Systems With Applications 200 (2022) 116941 4 DataPoint = {DayPoint i }, i = 1, 2, 3, ..., dd = TimeWindow (3) Where DayPoint is all features of one day, DataPoint is a data point. The time window d is varied as 3, 5, 10, 15, 30, 45 and 60 days. 4. Methodology 4.1. Wavelet transform (WT) WT has been widely used in image, speech, and signal processing. Many studies have shown that wavelet analysis is an effective denoising approach (Bruce et al., 2006; Kompella et al., 2016; Martínez and Gilabert, 2009; Masset, 2015; Patel et al., a. One of the main Table 3 Performance index of dichotomy. Indicators Formula Meaning Accuracy ATP+ TNT + F Correctly predict the proportion of samples in all samples Precision P = TP TP + FP The percentage of a sample that is predicted to be positive is also positive Recall R = TP TP + FN The percentage of all positive samples that were correctly predicted F score 2 F 1 = 1 P + 1 R Harmonic mean of Precision and recall TP: If an instance is a positive class and is predicted to be a positive class, it is True Positive. FP: If an instance is a positive class but is predicted to be a Negative class, it is False Negative. TN If an instance is a negative class but is predicted to be a positive class, it is False Positive. FN: If an instance is Negative, but is predicted to be Negative, it is True Negative. Download 2.42 Mb. Share with your friends:
|