missing values in Lists (vectors) and Lists of Lists (matrices)

[mittelmark]

Hello,

As far as I can see there is no support for inserting null into a list of doubles or integers, right? - Not sure if I understood those ?#! symbols correctly in the declaration.

My approach to indicate missing values would be to create classes for Vector and Matrix which keep track for the positions of missing values within a second List which contains the indices of missing values. Or may be a internal boolean List contains for every element true or false for indicating missing values. In my computations. I then have to look up in these parallel structures if there are missing values at a specific position.

Or are there other ideas to approach that typical problem?

Thanks!

[pfusik]

What you are looking for are double? and int? types. These are not implemented yet, but I want to add them soon.
As a workaround, perhaps you can use Math.NaN (for doubles) and int.MinValue ? Check NaNs with Math.IsNaN(value).

[pfusik]

BTW, see #31. Back in 2021, Fusion had no nullable references (all references were implicitly nullable). What still needs to be added are nullable numbers and enums (including bool?).

[mittelmark]

Thanks, that are some nice hints pointing to some possible solutions!

I'll check that out and as well how we can pass that to Swig, let's say how many languages we will loose. Null handling is always the killer .... ;)

But I think that is the last thing I need to start actually implementing my algorithm of interest.

I need to calculate Pearson and Spearman (rank based) correlations of two vectors, skipping missing values, then sorting my correlations values and comparing the order of correlations in two vectors. Let's see how that works out.