Fuzzy joiner

[KristianW1977]

Hi,

Beginner need some help.

I have two dataset from two different systems.
Systems are not fully syncronized and records have small differences and I need way to join to these records.

Differences between start time, end time and duration.
Mostly differences are in 10 second tolerance.

How I can combine/join records together?

Best regards
Kristian

[gonhaddock]

Hello @KristianW1977 and welcome to the KNIME community

You did mention about the 10 sec tolerance but you didn’t mention the the average duration of these periods. Or said it in question format: what percentage represents ± 10 sec over these periods?

I can assume as well that que are talking about an isolated tracking time system and not a multi parallel one. So the scenario is simple or multi systems are identified and can be looped through.

You can perform a Cross Joiner comparing pairs of possible vectors; and then calculate the overlap time length (or the average percentage of overlapped ) and the average percentage of unoverlapped tails. Many of these pairs won’t overlap, and assuming overlapping rates above (0.9?) will give you pairing match criteria.

I answered a topic some time ago related to systems overlapped time, it did identify starting-end timestamps of overlapping periods ; you may get some ideas for your workflow:

Based on system is there overlapping downtime

Hello, Thanks in advance for your help. System Overlapping Downtime.xlsx (11.7 KB) For my use case I would like to determine if the same system is down at the same time. As you can see I have three systems Earth, Wind and Fire. As per the highlighted lines Fire system is down on rows 4 and 7 and Earth system is down on lines 5 and 8. The time is not important but the overlap of the same date is important. Any help would be greatly appreciated. Kind regards, Sam

Some more literature about :

Identify overlapped items

Hi! I was wondering if there’s a way to identify overlapped values on two different columns. To give a bit of context, I have a table that looks somewhat like this [image] So, the thing is that for plant 123 the order B starts before the order A finishes and that type of overlapping is what I’m tryng to identify, but the data is huge and it would take my forever. Is there a way to make this easier?

Let us know about your thoughts.
BR