Convert Excel xls(x) to KNIME .table file format

KNIME Analytics Platform
1

Hello Forum,

perhaps this is a stupid question, but I spend some time on it with searching for knime workflows etc.

I would like to train a neural network (e.g. Keras, DL Phython etc.).
In the provided workflow examples I found on knime platform, always the Table Reader Node ist used.

My problem is, that I have my data in an excel file. My perhaps (silly) simple question is, how can I read an xls-file (xls-reader node) and convert it to a table format, which can be read then from the Network learner.

The sheet in the excelfile is pretty simple. It is only one sheet with name “Tabelle1” and in the first row I have named column A with the name INPUT1, column B with INPUT2 …column E with INPUT5 and column F with OUTPUT1.

In the Excel reader node “XLS Reader Settings” I have clicked Column names “Table contains column names in row numner 1”

From row 2 downwards it is containing data (only number,figures, no strings).

I tried the following workflow without success

Excel Reader Node (XLS) directly connected to Table Writer Node.
This simple workflow runs without warnings successfully.

If I wan tto use the generated xxx.table file and connect it e.g. to Keras Network Learner, I got the message

“The dialog cannot be opened for the following reason: Input table is missing or has no columns”.

Can somebody provide me with a simple workflow to solve this issue?

Many thanks upfront!

Bernd

2

Hi @Bernd

On the KnimeHub there is an Example workflow how the Excel Reader Node can be used. The output of this node is a Knime.table already.

If I understand you correctly your Keras example uses a Table Reader node as input. I’m not sure, but my guess is you could change this Table Reader node into the Excel Reader node. This avoids writing the table to disk and reading it again.

Have you checked the output of the Excel Reader node? Once it completed succesfully, right-click on the node and choose the bottom option. Is this as expected, or is it empty?

4 Likes
3

@JanDuo

many thanks - I found the mistake - the connection to the Keras Network Learner was wrong selected from my side. Now I struggle with the Keras Network Learner Node…
But this is another topic I have to solve…

Many thanks - your comment, that the excel reader node should work, was correct.

Bernd

3 Likes
4

Hi @Bernd,

any Reader nodes reads data in KNIME as KNIME data table. The Table Reader node reads file which was created with Table Writer node. Those two nodes use KNIME internal format which results in faster I/O. Hope this clarifies things a bit :wink:

Br,
Ivan

1 Like
5

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.