input_table._get_numeric_data() doesn't work in python predictor. What do I do to fix it?

[aysegulkocak1]

I am getting such an error in Python Predicator node while running my workflow.

The error is like that:

When I look at the previous node I see there is sampleID. It exists.

But I realized that problem is not about this.

This line in the code doesn’t work:

new_data = input_table._get_numeric_data()

So ._get_numeric_data() doesn’t work in this python code. It shows as Object type.

I am using Mac OS. Do you have any idea about solving the problem?

[steffen_KNIME]

Hi @aysegulkocak1,

Interesting.
Where does it show as object type?

What is the output of print(input_table['sampleID'])? If I have a column there which comes as an Integer column, I get:

Row0    1.0
Row1    2.0
Row2    3.0
Row3    NaN
Row4    5.0
Name: sampleID, dtype: float64

What is the difference between the output of print(input_table.columns) and the output of print(new_data.columns)?

Which KNIME Analytics Platform are you using?

Best regards
Steffen

[mlauber71]

@aysegulkocak1 could there be a problem with the index of a data frame. Maybe you can reset it.

Error in Python script node

Hi @mohini1329, as the others have already suggested, you should make sure that each row in the Pandas DataFrame has a unique index. Dropping/resetting the index is not the best option though, because KNIME uses the RowID (which is equal to the DataFrame index) to identify rows when you select / highlight individual cells in your data. If you reset the index, then this row identification will no longer work. The best solution would be to let the index of all values in your input table as they…

[aysegulkocak1]

Hello,

Thank you for your answer.

The output of print(input_table[‘sampleID’]) is :

The difference between the output of print(input_table.columns) and the output of print(new_data.columns):

And I’m using KNIME 4.7.2.
Thanks.

Best,
Ayşegül Koçak

[steffen_KNIME]

Hi @aysegulkocak1,

thanks for the details. Strange, I cannot reproduce this.

1. Can you provide your pandas version?

   import sys, pandas
   print(pandas.__version__)
   

2. Could you try resetting the index as suggested (and explained how to do it) in the post linked above by mlauber?

3. And, lastly, could you provide a minimal workflow to reproduce this issue?

Best regards
Steffen

[aysegulkocak1]

Hi,

Thank you so much for your answer again.
The version is 2.0.1.

Actually, I tried resetting the index but it didn’t solve the problem

Okay, I will try it with minimal workflow.

Thanks so much.
Best,
Ayşegül Koçak

[Daniel_Weikert]

if possible upload a sample workflow with demo data
br