Handling wrong or missing data

[fhenneke]

If I read the code correctly, the code does not contain utilities for handling missing values or recomputing imbalances for hashed when data is faulty.

It would be good to add such functionality so as to be able to estimate the resulting code complexity and workflows related to this experiment.