ML awareness

Confidence levels are a key concept in everything we do. See https://contractfit.atlassian.net/wiki/spaces/DA/pages/347078920 for a more detailed explanation. Essentially, confidence levels are a key prerequisite to automation and Exception-based review .

It is very important that we have a strong correlation between the confidence levels and correctness of a prediction. We colloquially refer to this correlation as the "Awareness" of our solution. We offer different charts to provide you with visibility on awareness.

  • Awareness by field

 

Field awareness

In this graph, you have one dot per prediction. The value in the y-axis shows the confidence with which the prediction was made, the color shows whether the prediction was correct or not. This is an important indicator for automation: high confidence predictions should be correct most of the time for you if you want to be able to automate with a low error rate.

 

Related pages