Subject: | Re: Fuzzy logic search for text matching
| Date: | 13 Jun 2022 18:37:24 -0400
| From: | "Jeremy Stagg" <jeremy.stagg@itsysadmin.com.au>
| Newsgroups: | pnews.paradox-programming
|
Thank you Kevin
Sage advice.
Cheers
Jeremy.
"Kevin Zawicki" <numberjack@wi.rr.com> wrote:
>
>But then why not clean up table 1 and change custname to in table 2 to match
>table 1 based on Cust Code to start?
>Then you know all custcodes have same name in both tables.
>Then if you might have a cust name in both with different cust code, find
>all cust codes in each table that are not in the other?
>
>Or maybe
>1 00301 MOOROOLBARK SECONDARY COLLEGE
>2 00301 MOOROOLBARK COLLEGE
>
>I would have tcursors scan the CustName column and break cust name into
words
>on spaces, and put in another table, doing this on both tables into one
table
>
>Seqnum | tablenum | custcode | word |
>---------------------------------------
>1 1 00301 MOOROOLBARK
>2 1 00301 SECONDARY
>3 1 00301 COLLEGE
>4 2 00301 MOOROOLBARK
>5 2 00301 COLLEGE
>
>Then do some sum counts on common words by cust codes.
>
>My experience is that you cannot full automate this, you have to use the
>code and analyze to find, then edit / cleanup
>
>
>
>
>
>"Jeremy Stagg" <jeremy.stagg@itsysadmin.com.au> wrote:
>>
>>Hi Kevin
>>
>>Both databases have the same Customer table structure and keyed:
>>
>># | Field Name | Type | Size | Key
>>----------------------------------
>>1 Cust Code A 5 *
>>2 Cust Name A 35
>>
>>There are discrepancies between the two tables / databases by Cust Code
>and
>>then Cust Name.
>>
>>Examples are:
>>
>>Table | Cust Code | Cust Name
>>-----------------------------------------
>>1 00301 MOOROOLBARK SECONDARY COLLEGE
>>2 00301 MOOROOLBARK COLLEGE
>>1 08415 BACCHUS MARSH COLLEGE
>>2 08415 BACCHUS MARSH SECONDARY COLLEGE
>>1 07179 BAKER MEDICAL RESEARCH INSTITUTE
>>2 07179 BAKER IDI HEART & DIABETES INSTITUTE HOLDINGS
>>
>>There are % similarities to identify and grade, allowing for a quick remediation
>>and then can consolidate the data sets.
>>
>>The Cust Code is the pinacle key as there are more structures stemming
from
>>this, such as Customer Address.
>>
>>I'm making Table 1 the Gold Master and making changes only in Table 2.
>>
>>Hence the fuzzy logic.
>>
>>Have used the Google Sheets solution from:
>>https://stackoverflow.com/questions/70989502/fuzzy-matching-in-google-sheets
>>
>>And looking at this it still says to me Paradox could do it.
>>
>>
>>"Kevin Zawicki" <numberjack@wi.rr.com> wrote:
>>>
>>>Can you be more specific?
>>>Are they keyed?
>>>Two tables with same customer key # or name or anything?
>>>If the two tables have any data field the same and you are looking for
>record
>>>level differences you could join the two tables and check all and use
!,
>>>something like that.
>>>Need more info…
>>>
>>>
>>>
>>>
>>>"Jeremy Stagg" <jeremy.stagg@itsysadmin.com.au> wrote:
>>>>
>>>>Hi to All
>>>>Wondering if anyone has scripted in ObjectPAL fuzzy logic for identifying
>>>>text matches in columns between two tables?
>>>>I have to identify matches between two separate tables of customers (separate
>>>>databases) that have "drifted" and looking to merge the two.
>>>>Cheers
>>>>Jeremy.
>>>
>>
>
|