Invalid training data. Labels must not contain undefined values.

Trying to train an LSTM with 3 classes, and 11 features. trainX is a 507*1 cell array of 11*n double matricies, and trainY is a 507*1 cell array of 1*n categorical vectors.

Here's my code

optsX = detectImportOptions('Ordered Name.csv');
optsX.SelectedVariableNames = [18,25,26,32,33,34,35,36,78,80,81];
holdX = readmatrix('Ordered Name.csv',optsX);

optsY = detectImportOptions('Ordered Name.csv');
optsY.SelectedVariableNames = 1;

holdY = readmatrix('Ordered Name.csv',optsY);

types = {'A','B','C'};

top = 1;
previous = 1;
counter = 1;
trainX = {};
trainY = {};
for i = 1:size(holdX,1)
    if(previous - holdX(i,9) > 1)
        trainX{counter,1} = holdX(top:i-1,:)';
        trainY{counter,1} = categorical(holdY(top:i-1,:)',types,'Ordinal',true);
        top = i;
        counter = counter + 1;
    end
    previous = holdX(i,9);
end
trainX{counter,1} = holdX(top:i-1,:)';
trainY{counter,1} = categorical(holdY(top:i-1,:)',types,'Ordinal',true);

numFeatures = 11;
numHiddenUnits = 200;
numClasses = 3;

layers = [ ...
    sequenceInputLayer(numFeatures)
    lstmLayer(numHiddenUnits,'OutputMode','sequence')
    fullyConnectedLayer(numClasses)
    softmaxLayer
    classificationLayer];

options = trainingOptions('adam', ...
    'MaxEpochs',10, ...
    'GradientThreshold',2, ...
    'Verbose',0, ...
    'Plots','training-progress');

net = trainNetwork(trainX,trainY,layers,options);

I keep getting the error: Invalid training data. Labels must not contain undefined values. I was unable to find any sort of question that had this error answered, hoping for someone pointing out something stupid I missed

Edit: Names modified to be more general

Edit 2: I'm a big dumb who didn't notice that some of the labels listed in his data were null. Leaving this up for someone else who encounters this to check their data.

[Walter Roberson]

trainY{counter,1} = categorical(holdY(top:i-1,:)',types,'Ordinal',true);

You are building a categorical. If any of the values selected out of holdY are not in the list indicated by types, then you would get an <undefined> in the categorical result.

[NICHOLAS WILKE]

Yep, see my Edit 2. That's exactly what happened.

[ASHLY JOSE]

Hi, Iam facing the same problem, but iam dealing with images. what do you mean when you say 'some of the labels were null'?

Thank you

[Walter Roberson]

Some of entries in the data file were empty instead of having a valid class