Machine Learning use cases - Audit

[Farriyajawed]

Hi everyone, 

I have recently started using Intl. suite and I was wondering what Machine Leaning or intelligence suite tools are most commonly used in Audit Procedures. 

Any use cases would be helpful!

[acarter881]

Hey @Farriyajawed, my guess is the vast majority of audit procedures would not use any tools from the Intelligence Suite. Do you have any specific audit procedures in mind?

[aatalai]

@Farriyajawed I would say the computer vision tools might be useful if you have pdfs in a consitant format, that way you can scan them into alteryx fairly quickly without having to manually type them in.

If you have journals or something with fraud flag from previous years you might be able to use them to predict journals that might be at a higher risk of being fraud/incorrect

[c-lopez]

Hi @Farriyajawed,

I've worked with some customers on predictive use cases for audit, I hope I can shed some light on this topic. Firstly, I am glad to hear you've started using Intelligence Suite (IS)! I am going to answer with the assumption that you are referring to the ML tool in IS rather than standalone Alteryx ML in the cloud. 

I would say that this list is pretty accurate to what we have seen before as far as applicable use cases:
https://www.alteryx.com/resources/e-book/5-use-cases-to-help-audit-professionals-take-back-their-tim...

But as far as the most useful tools, it really depends on your data and most importantly what is going to matter to you in terms of ROI or value, I always like to say that value is king. With that said the most common use of Intelligence Suite at the beginning is using the computer vision and text mining tools to get your data on the proper shape to do more advanced analytics (predictive). If you think about it they are an extension of the other Alteryx tools to help you extract and shape data. 

For example, one of the use cases I have heard about before is using IS computer vision tools to extract data from PDFs submitted by employees submitting business expenses and using predictive tools downstream to find the likelihood of the expenses being fraudulent or not using a classification model. 

Really sky is the limit, if you think of a use case of your own please let us know!