urbste
commented
12 hours ago For this please check this paper. Also check Figure 6 and Equation 13.
In the end it is a hyperparameter and you have to find one that works good for your data.
To select a suitable knot spacing, ∆t, we thus first decide on a quality value, ˆ q ∈ (0, 1], that corresponds to the fraction of signal energy we want the approximation to retain.
@urbste Hi author, Thank you for share your project, How should I determine the sensor quality parameters of the get_sew_for_dataset.py. I found that different parameters produce different calibration results because they affect the weights in the final spline optimization process.