Stickleback generates local event probabilities by sliding a prediction window over the longitudinal data. For efficiency, predictions are only generated for every nth window, where nth is set in Stickleback.init(). I think it makes more sense to have a percent overlap instead, which is converted to nth. E.g., if window_size is 100 and percent_overlap is 0.5, then nth is 50.
Stickleback generates local event probabilities by sliding a prediction window over the longitudinal data. For efficiency, predictions are only generated for every nth window, where nth is set in Stickleback.init(). I think it makes more sense to have a percent overlap instead, which is converted to nth. E.g., if window_size is 100 and percent_overlap is 0.5, then nth is 50.