You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thanks for implementing this great package.
As I understand, the algorithm should find the elbow/knee where the maximum y_difference is.
However, when I test this algorithm is in 'online' mode (for an increasing convex curve), it tends to find the elbow where the last local maximum y_difference is. (shown in the figure below). Since the original paper does go into details about the online correction, I wonder if this behaviour complies with the original algorithm.
The text was updated successfully, but these errors were encountered:
Without the data and/or code provided, it's hard to help you with that. For the online setting, have you used a sensitivity of 1, as in the paper suggested?
Hi,
Thanks for implementing this great package.
As I understand, the algorithm should find the elbow/knee where the maximum y_difference is.
However, when I test this algorithm is in 'online' mode (for an increasing convex curve), it tends to find the elbow where the last local maximum y_difference is. (shown in the figure below). Since the original paper does go into details about the online correction, I wonder if this behaviour complies with the original algorithm.
The text was updated successfully, but these errors were encountered: