Well, an answer might depend on some additional goals (objectives).

For example,

if a goal is the "minimal L2-norm of the second derivative of the interpolant in a strip"

then the answer / the solution is "a cubic spline" ... according to

Article Best Interpolation in a Strip

When selecting a suitable spline for modeling non-linear data, there are several factors to consider:

Complexity of the Model: Depending on the complexity of the data, the choice of spline type may vary. For example, if the data has a simple and smooth structure, a natural cubic spline may be sufficient. On the other hand, if the data has more complex, non-linear structures, a more flexible spline like a B-spline or penalized spline may be required.

Degree of Smoothness: The degree of smoothness required for the model is also an important consideration. Natural cubic splines tend to produce smooth curves, while B-splines and penalized splines can be used to model more complex, rougher curves.

Number of Knots: The number of knots required to model the data is also important. Natural cubic splines require a fixed number of knots, while B-splines and penalized splines can use a variable number of knots. This can be advantageous if the data has a large number of features, as it allows the model to adapt to the data more flexibly.

Overfitting: Overfitting occurs when the model is too complex for the data, and it begins to fit to the noise rather than the underlying structure of the data. Penalized splines can help to prevent overfitting by imposing a penalty on the size of the coefficients used to fit the data.

In summary, when selecting a suitable spline for modeling non-linear data, it is important to consider the complexity of the data, the degree of smoothness required, the number of knots needed, and the potential for overfitting. Natural cubic splines are a good choice for simple, smooth data, while B-splines and penalized splines may be better suited for more complex, non-linear data