Table 3.
Features and Feature Extraction Methods.
| Ref. | Features | Extraction Features in Segments/Whole Session (Time) 1 |
Feature Extraction Methods |
|---|---|---|---|
| [52] | Standardized SCR & SCL score |
Time segment around event (20 s) | Ledalab |
| [53] | Mean | Task segment (varying) | - |
| [54] | Mean | Task segment (varying) | Manual |
| [55] | Standardized SCL score | Time segment (2 min) | Biograph Infiniti |
| [39] | Mean | Whole learning session (45–60 min) | Manual |
| [56] | Mean, SD, min, max, percentiles | Time segment (1 min) | cvxEDA-tool |
| [57] | Mean, SD, min, max | Time segment around event (90 s) | - |
| [58] | Mean | Time segment (1 min) | Ledalab |
| [49] | Mean | Task segment (40 s) | Ledalab |
| [59] | Mean | Whole learning session (2 h) | Manual |
| [60] | Standardized SCL score | Time segment around event (5 s) | Ledalab |
| [46] | Mean, range | Time segment around event (10 s) | Augsburg toolbox |
| [61] | Number of SCR peaks, Standardized SCL score | Whole learning session (2.5 h) | - |
| [40] | Mean | Task segment (varying) | - |
| [62] | - | Time segment (10 s) | Augsburg toolbox |
| [63] | Mean | Time segment (1 min) | - |
| [36] | Mean | Whole learning session (-) | Ledalab |
| [50] | Mean | Task segment (-) | Neurokit |
| [64] | Number of SCR peaks, Frequency of SCR peaks | Time segment (1 min) | Ledalab |
| [65] | Mean | Task segment (4 min) | - |
| [66] | Amplitude sum of SCR peaks, Latency of SCR peaks | Whole learning session (1 h) | Ledalab |
| [33] | Number of SCR peaks, Onset of SCR peaks | Time segment (1 min) | Ledalab |
| [67] | Mean | Task segment (varying) | - |
| [68] | Frequency of SCR peaks | Time segment (1 min) | Ledalab |
| [69] | Mean, Number of SCR peaks | Task segment (59–79 s) | Acqknowledge |
| [70] | Mean | Whole learning session (75 min) | Manual |
| [71] | Mean | Whole learning session (-) | - |
- means no information is given. 1 Extraction of features from the EDA signal was done in segments or over the whole learning session. Task segments are based on the time spent on a task. Time segments are specific periods of time, which also can be initiated around a specific event (such as entering an answer). Whole learning session: EDA features are extracted from the whole track, which consists of multiple tasks.