Naive Bayes Classifier in SDI

Hi All,

I am trying to replicate the Naive Bayes scenario from the documentation using SDI, just to get the hang of how it works. So I began by creating the training model, and that worked perfectly fine. I got the same results as in the documentation, so that's good.

The problem comes when I try to create the prediction model. Here are the steps I followed:

1. Create the table for the prediction input data (PAL_NBCPREDICT_PREDICTION_TBL in the doc)
2. Create the table for the prediction model (same as NBC_PAL_NBC_MODEL_TBL, and this is populated by the training model)
3. Create a new SDI flograph with the NBCPREDICT node, connect the input table and model table to the DATA and MODEL elements within NBCPREDICT

Now here, it throws an error saying that the table structure is incorrect for MODEL and PARAMS tables within NBCPREDICT. This is strange as the MODEL table has the same structure as the output of NBCTRAIN. The MODEL.ID type is integer, and that's what is used in the doc as well. And I didn't even change the PARAMS table structure, it is exactly the same as in the doc, yet it tells me the structure is wrong.

I went one step further, and modified the model. Now, I placed both NBCTRAIN and NBCPREDICT in the same model and directly connected the output of NBCTRAIN --> NBCPREDICT MODEL table. See below. I still get the same error message.

Here's what the MODEL and PARAMS table structures look like - they're exactly as in the doc.

And this is the part where I'm stuck - it simply doesn't accept the input tables I provided, yet the data types seem to be exactly as stated in the doc. Would appreciate any help or pointers on this.