P1:
I started out with an individual case. 
So, report comes in some kind of model says that it's most likely fraud. 
Then a fine gets sent by mail. 
And then with that fine comes some kind of way to go to an inference dashboard or whatever I would call it. 
Where on an individual case level you can see all the indicators that went into the decision basically. 
And then you have a potential for appeal based on the inference dashboard basically.

P2:
Yeah.

P1:
So, what I was lacking and what inspired me from the infographic was the monitoring part because I think these decisions can be steered, can be helped to steer policymaking. 
So, what the most important part I think is some kind of monitoring dashboard on an aggregation scale. 
So, not individual cases but on an aggregation of all outcomes of the models that is both visible for policy makers but also for the greater public. 
So, my aggregation dashboard has three parts which is overall statistics. 
Because I always feel like that is not taken into account. 
Like, most specifically the amount of days that are just normal days that go well and then the amount of days that are fraudulous. 
So, that there is always a good overview of what we're talking about and what the context is. 
So, let's say 99% goes well but that 1% is our fraud cases and over time obviously if it increases it will go down because it can also be used to change policy.
Then I have a section on contributors.
So, the more or less SHAP value based which features go into the system and what makes an impact.
Specifically, I have added some statements that you sort of hopefully can generate for example so that can be helpful for for policy making too. 
I put some examples in for example in the city centre there's more fraud cases but also one that I put in is older owners have a bigger chance of breaking the law which could be indicated that the law doesn't reach older citizens for example and then last part I ran out of time but I wanted to make an overview of some bias like the model drift over time and stuff like that. 
So, that was my idea.