*5.3. Time Intervals*

The analysis was repeated for different intervals within the bursts to investigate the intervals in which classes may be identified in the *Swift*/BAT sample.

First, the feature extraction analysis with 4 ms resolution light curves was performed for the interval of T<sup>0</sup> − 1 s to T<sup>0</sup> + 1 s (Figure 8a). The addition of pre-trigger data is shown to produce almost identical results to those obtained by starting at T0. However, the analysis requires additional Principal Components to explain the variance, indicating that including 1 s of data before the trigger adds more noise than information. Secondly, when the selected interval is between T<sup>0</sup> + 1.004 s and T<sup>0</sup> + 2.008 s, the separation disappears, as shown in Figure 8b, indicating that the early prompt emission in the first second post-trigger is the key interval for separating the two classes.

**Figure 8.** 2D t-SNE representation of the wavelet feature extraction applied to *Swift*/BAT light curves covering time intervals (**a**) T<sup>0</sup> − 1 s to T<sup>0</sup> + 1 s and (**b**) T<sup>0</sup> + 1 s to T<sup>0</sup> + 2 s. The plots are coloured by burst duration T90.
