PCA Apply Node not working

Hello anyone that can help here, it would be much appreciated.

I saved the PCA model during training and when I apply it to new data for inference I am getting this error

WARN PCA Apply 7:668:642 The model is expecting column “definit” which is missing in the input table

Up until now, I had a train pipeline, where I use TF-IDF for text data, normalize it, and then apply PCA. I save the PCA model and then apply to new data to make predictions.

Now suddenly its giving me this error that it doesn’t see this specific word. This should predict on new data and obviously text will never be exactly the same.

Thanks!

Is the column in fact missing from your test data?

If you trained a model using data that certain defined properties and your test data doesn’t have those properties, then the model can’t be applied.

1 Like