Prediction using Train and Test datasets

KNIME Analytics Platform
1

Hello. I have two text files, one is train dataset and other is test dataset. I want prediction but have no idea how to predict the test dataset based on train dataset? I am using KNIME 3.4. 

 

Best Regards

Alam

2

Hi Alam,

give our learning material on our website a try: https://www.knime.com/nodeguide/analytics. There (and on the example-server within the KNIME Analytics Platform) you will find many examples on how to build a predictive model. If you still have questions after studying these materials, don't hesitate to ask me again.

Cheers,
Marten

3

Thank you very much Marten. I am attaching my workflow. Using Naive Bayes Predictor the results of prediction are seems wrong.

Thanks

Alam 

4

Hi Alam,

why do you think the results are wrong and what does it even mean in this classification task context?

Cheers
Marten

5

Hi Marten 

Yes the results are wrong. In this task i have train dataset and i am test datasetht on it to find the accuracy and prediction. When i am running the predictor give me 0% accuracy.

Regards

Alam

6

Your test dataset does not even have labels ("?"), so how would you determine an accuracy for it?

What's the idea behind this task?

7

Hi qqilihq

Oh yes, you are right :) actually i want to predict the accuracy of my test dataset. 

Regards

 

Alam

8

Ok (I assume you mixed up test and train in above's post again, but nevermind ?)

As I'm on a mobile currently, I cannot look at your workflow. What kind of preprocessing do you apply to the text before feeding it to the Naive Bayes nodes?

9

Hi qqilihq.

I have done "Preprocessing" (Row Filter, Column Filter, String to Doc, Case converter, Stopwords and Stemming)

Regards

Alam