CODEWITHSUNDEEP
pythonmatrixmissing-datafancyimpute
JPiter
JPiter

Reputation: 91

Matrix Imputation with NuclearNormMinimization fails with TypeError

I am trying to impute missing values in the matrix by calling different functions from fancyimpute.py. NuclearNormMinimization runs fine for some of the matrices but fails for the example below.

NuclearNormMinimization(verbose=False).complete(matrix)

where ```

matrix = [[3.64625, 7.34375, 15.0], 
         [5.00000, 12.50000, NaN],
         [15.00000,11.25000, NaN]]

The error message is:

TypeError: Expected NuclearNormMinimization.solve() to return NumPy array but got class 'NoneType'.

I am using this approach for bigger sparse matrices, but once in a while I get a small matrix with a few missing values, and my code fails. Can somebody help me to understand why this is happening? What are assumptions for NuclearNormMinimization imputation techniques?

Upvotes: 0

Views: 295

Answers (0)

Related Questions