'Explain' feature in OAC

Question

Summary

Reliability of OAC 'Explain' Feature

Content

Hello,

I've been Exploring OAC's "Explain" Feature to generate meaningful insights out of my data. For ex, I have ErrorCode as one of the fields in my data and I would like to find the probability of occurance of a particular ErrorCode based on other fields in my data, such as, Operation, Application Type, Application Version and so on.

To achieve this, I did right click on the field "Application Type", in my data visualization and selected "Explain", it gave me an insight that, I could interpret (attached here). This is really good.

Now the Problem I am facing is, this feature is inconsistent. Everytime, I click on the same field and say, "Explain", it does not show me the insights, as expected. Instead, it sometimes pops up a message saying, “Something Went Wrong". Sometimes, it errors saying, "There was no data that was strongly correlated to outcomes for APPLICATION_TYPE All Values".

Hence, I want to check in this forum, the reliability of this feature in OAC. Also, any tips/user experience, if some one has already been using this.

Thanks in advance,

Chethana

Chethana Kumari-Oracle · Answer

Hi Jeffy,

Appreciate your response. Thank you so much.

Regards,

Chethana

Jeffy Joseph · Answer

Hi Chethana,

I have used the explain feature as a part of my descriptive analysis often. However, I haven't received this error yet.

One tip that I can give you regarding the explain parameter is that apart from identifying the key drivers behind an attribute or a measure, it also sometimes suggests anomalies based on your data points. So, this can help you get a sense of outliers in your data as it can compare actual values vs projected values. In case you're planning to do some machine learning on top of your data, it can be a useful feature as it helps you omit those points while training your machine learning model.