How to get data from inaccurate OCR (inconsistent line feed and number mistaken as alphabet)?

[SunnySchindler]

For scanned text from statement of credit card transactions, how to use BI to get data without pre-processing.

Scanner output (text file) has following attibutes:

1. partially accurate OCR
   1. same line can have more than one transaction
   2. some number became text
2. each transaction
   1. start with date (DD MMM)
   2. description
   3. Location
   4. reference ID
   5. currency amount if applicable
   6. Amount converted to my currency
   7. Amount may have text suffixed (such as credit)

Do I have to pre-process before using Power BI?

Kind Regards

P.S.

Microsoft Power Automate does not yet have a template for credit card statement and requires admin authorisation to customize AI model built for invoice template.

[Anonymous]

Hi @SunnySchindler ,

In Power Query, you can use split column feature to extract your value.

Extract Parts of a Text Value in Power BI using a Delimiter: Power Query Transformation - RADACAD

Best Regards,

Stephen Tao

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.