Predictions
Which running cases will miss the deadline - flagged days before it happens.
This screen does not exist yet. Nothing on this screen is computed and there is no data behind it. What follows is a description of the screen, written so the shape of the product can be read - it is not a result. The only figures on this page are the container's own event and case totals in the bar above; nothing below them has been measured, because nothing has been taken.
What this screen will do
This screen will score the cases still running against what this project's own history says about cases like them, and put the ones heading for trouble at the top.
What it will show when it exists
- At-risk cases sorted by risk, each with a risk meter and each drilling into its own timeline.
- A box stating how good the model actually is, in plain terms, which does not shrink when the answer is unflattering.
What it will need first
A model trained on this project's history, and cases still running to score. Neither exists yet, so there is no risk to report here - not a low risk, none at all.