Solved

How to check duplicate records in ONE Desktop, and using fuzzy logic?

  • 22 May 2023
  • 4 replies
  • 226 views

Userlevel 1
Badge +1

Hello. i have a general material data csv file, and in that i have to check for duplicates based on material data. Currently i'm trying with the matching step by creating partitions and key rules but still i'm not sure about it. So is there anyone knows about solving this process?? Thanks!

icon

Best answer by AKislyakov 25 May 2023, 10:23

View original

4 replies

Userlevel 4
Badge +4

Hi Suryakanth,

You can find a tutorial on how to configure the Matching step in the Tutorials project > 09 Match and merge > 09.01 Match and merge.plan.

If you're new to this, I suggest trying the Canopy Clustering step first. It requires less configuration and is more user-friendly. You can find an example usage in the Tutorials project > 07 Analyze > 07.10 Clustering.plan.

The Canopy Clustering step allows you to setup components which will be used to search for duplicates, their weights and thresholds when to consider two records belonging to the same group.

 

 

Userlevel 1
Badge +1

Hello @AKislyakov ,I'll work on this method and let you know the resultant, thanks.

Badge

Hi @AKislyakov, I am following the guide on this tutorial however, it’s quite hard for me to understand each functionality although I used the help button to understand what is it all about. What I am looking for is to detect duplicates on several columns like Name, Address, etc. After following the guide, I found that my result turned up to a be a bit confusing whereby different Name values belong to one same cluster. For example Yayasan A and Kementerian in one cluster ID while I am expecting Name like Yayasan A, Yysn, Yayasn A belong to the same cluster and Kementerian belong to a different cluster ID. Where should I change in the Canopy Clustering the configuration to get the correct result?

Userlevel 6
Badge +7

Hi @Radziah, thanks for posting! Have you had a chance to check any resources on the Jaccard Index? If you haven’t, I would highly suggest checking some such as this, for more perspective on clustering that can possibly clarify the results. If you need additional support please feel free to raise a support ticket with screenshots at support.ataccama.com and our team would be happy to help 🙋🏻‍♀️

Reply