Duplicate Records Inspection - Help to find them?

[craig157]

Hi All,

We have a couple hundred thousand customer records on our system as the moment and I was hoping to investigate what can be merged then deleted to keep our record count down.

I will be trying to check this over several different feeds as unfortunately most of these fields can be manually written so will have different formatting mistakes & spelling etc.

My data is like below:

OrgID	Company Name	Postcode
1	Blahblah	AB14 00T
2	Bladeblah	AB14 00T
3	RunningoutofNames	AB12 111
4	MadeupName	AZ4 78T
5	Notsharing	TR4 9E9
6	PenguinDance	XER 101

What I'd like to be able to do is Identify Companies with the same postcode that share a certain number of characters.

So Orgid 1 & 2 would match as they share 8 of the same characters in their company name.

Ideally the table would only show OrgID 1 & 2 now as the rest don't fit the criteria.

Anyone got any ideas on how I would enable this to display?

Please note - there may be more than one instance where a company name is duplicated under the same postcode.

Thank you,

[Anonymous · Accepted solution]

maybe like this:

(you donot Need to drop the origin table, but you need to use other fieldnames)

TEST:
load *  inline [
OrgID,Company Name,Postcode
1, Blahblah, AB14 00T
2, Bladeblah, AB14 00T
3, RunningoutofNames, AB12 111
4, MadeupName, AZ4 78T
5, Notsharing, TR4 9E9
6, PenguinDance, XER 101
];

left join(TEST)
load count(OrgID) as Numb,
Postcode
Resident TEST
group by Postcode;

Final:
NoConcatenate load *
Resident TEST
where Numb > 1;

drop table TEST;

[jmvilaplanap]

Hi,

You can use fuzzy logic to compare strings, I found this post. Maybe will be useful

Approximate String Matching in QlikView

[MarcoWedel]

maybe helpful:

String Percentage Comparison?

regards

Marco