UnboundLocalError: local variable 'X_val_nots' referenced before assignment.

[imb-ai-junghwan]

Hi,

I have an issue using GDmicro.

My dataset has 16 samples [healthy; 7 and disease; 9] of train datasets and 9 samples [healthy;4 and disease; 5] of test datasets. (SIAMCAT error had not occurred)
Due to the dataset size was too small to perform 10-fold cross-validation, I set close_cv == 1.
The test_idx length is 9, and the parameters are doadpt = 1, reverse = 0, and uf = 1.
Since len(test_idx) < 12, reverse = 0, and uf = 1, the build_graph_mlp function in run_MLP_embedding is executed.
However, because close_cv = 1, the variable "X_val_nots" is not defined.

As a result, the code _, _, output = mlpc_raw(X_val_nots) raises an error:
UnboundLocalError: local variable 'X_val_nots' referenced before assignment.

How can this issue be resolved?

Thank you in advance.

[liaoherui]

Hi, I apologize for the issue and the delayed response...

I suspect this is due to a SIAMCAT limitation: if I recall correctly, it requires at least >10 samples per class (healthy/disease). The small sample size here may therefore cause problems.