Methods for Parsing Semi-Consistent Data

[Anonymous]

I'm looking to model my personal banking transactions. I'd like to categorize each transaction into 'Groceries', 'Restaurant', etc. so I can see where the money is going from at a higher level. 

From a data anlysis best practice point of view, how should I go about parsing that data? Are there any tools that are currently available? Any known methods for handling this kind of task? Would it be possible to use some kind of data analytics language to look for specific keywords and come up with a score?

For example, some descriptions look like below....

TARGET        00095830

SHELL OIL 85326124306

GRUBHUBSOMERESTAURANT

That would then be categorized as...

Grocery

Auto

Restaurant

Thanks in advance!

[tringuyenminh92]

Hi @Anonymous,

I think you could use Bing spell check to correct your mixing text and Text Analytic - Topic dectection to recognize the category of description as short-term solution. But it's long story cause we could not ensure it's correct 100%. Beside applying these service, you could build up and train LUIS model to understand your daily data, so it will grow up day by day to achieve what you are expecting.