I’m attempting to calculate the embedding dimension using FNN from nolitsa.dimension.fnn function. My time series is force on left leg during gait analysis.

dim = np.arange(1, max_dim)
f1, f2, f3 = dimension.fnn(force_l_leg, tau=time_delay_l, dim=dim)

Using this code, I keep getting the following error:

Exception: Could not find any near neighbor with a nonzero distance. Try increasing the value of maxnum.

max_dim (301) is defined as 10% of the length of the left leg force array (3015), and the time_delay_l for this specific calculating is 15, as calculated via Average Mutual Information method.

Wondering if anyone has any experience in this area and knows how to trouble shoot this error. Specifically, if there’s a problem with my inputs, or if a relatively cyclic array such as gait analysis is inherently prone to errors when calculating embedding dimension using FNN.

I was attempting to calculate embedding_dimension using nolitsa.dimension.fnn but unfortunately am running into errors. I attempted to manually input smaller and larger values for both tau and max_dim but have no luck.

For context this is what the Force L Leg looks like: