pmml4s model.predict() returns array instead of single value

I used sklearn2pmml to serialize my decision tree classifier to a pmml file. I used pmml4s in java to deserialize the model and use it to predict.

Iuse the code below to make a prediction over a single incoming value. This should return either 0/1/2/3/4/5/6.

Object[] result = model.predict(new String[]{"220"});

The result array looks like this after the prediction:

Does anyone know why this is happening? Is my way of inputting the prediction value wrong or is something wrong in the serialization/deserialization?

It is certainty of model for each class. In your case it means that it's 4 with probability 94.5% or 5 with probability 5.5% In simple case, if you want to receive value, you should pick index for the maximal value.

However you might use this probabilities to additional control logic, like thresholding when decision is ambiguous (two values with probability ~0.4, etc.)