Grouping along data pattern

Hi everyone, 

first of all thanks for all the helpful contributions here. I have learned a lot from previous problems. But here is one I couldn´t solve yet. 

I want to group rows along repeating pattern... example

City Path Time Session
New York www... 2017.08.31 1
New York www... 2017.08.31 1
Madrid www... 2017.09.01 2
Barcelona www... 2017.09.01 1
Barcelona www... 2017.09.01 1

For the previous table I would love to get 3 grouped rows. The first one containing row 1+2, the second containing row 3 and the third containing 4+5

I have nearly 500.000 rows in my dataset. The pattern is indicating the sessions of web users. But there is no unique session ID. If there where, I would use the Group by node. I tryed the Group Loop but its complaining that the ID is repeating.

So far I am out of options! ;-) 

Maybe I need to generate a session ID first. Is there any suitable cluster procedure what you could recomend?

 

Thanks in advance

Lars

 

 

 

Hi there,

 

I think all you need is the GroupBy node.

You'll want to have the group column as City, and in the 'Manual Aggregation' tab, put Path, Time and Session and make sure the aggregation type is 'First'.

It sounds like you want the same order of the rows, so in the advanced settings, tick 'Retain row order'. Also you will probably want the same column names so in the 'Column naming' drop down box, select 'Keep original name(s)'.

 

Let us know if this works for you.

Hi there, 

thanks for the tipp. I will give that a try - right away. 

So far, I have helped myself with an artifical session ID. And then used the "group by" node on the session ID. Its not 100% accurate but 99%. I found one case out of 500.000 that was wrong. I can live with it.

I will give an update if your suggestion worked!! 

Cheers

Lars

 

 

 

 

Thanks again for the hint. It didn´t worked 100%.

However, the problem is solved with the session ID.

 

Cheers

Lars