I am attempting to calculate Kendall's tau for a large matrix of data stored in a Pandas dataframe. Using the corr function, with method='kendall', I am receiving NaN for a row that has only one value (repeated for the length of the array). Is there a way to resolve it? The same issue happened with Spearman's correlation as well, presumably because Python doesn't know how to rank an array that has a single repeated value, which leaves me with Pearson's correlation -- which I am hesitant to use due to its normality and linearity assumptions.
Any advice is greatly appreciated!