Question on inspection

Jeff-HOU

Hi,

I have two questions regarding inspection:

1. Does anyone have any ideas on the purpose of inspection, other than penalty_V5 which reflects the result of "on_site" inspection. Is it just for checking if the actual area matches with "area" attribute? Or may it lead to some penalties?
2. Can we use event "inspection" to work on Q1? Q1 wants us to detect untimely cases "as early as possible". However, since we have no idea on when the derived attributes are set for these cases, I just wonder which attributes we can use to do the detection.

Thank you very much!

Regards,
Jeff

fborchert

1. Yes, penalties can be the result of an inspection. However, penalties may be the result of other control activities (such as reference alignment).

This is the at the core of Q2 (risk assessment):
How do you pick a sample for inspections, for which you are likely to find many severe penalties, that you would otherwise miss or find only later by other means.

2. Q1 (undesired outcomes) is not particularly linked to inspections. The detection should be made before the first occurrence of Payment application+Application+decide.
This usually happens much later in the year than the inspections, so yes, you may use all events relating to inspections.
You may also use all derived attributes with suffix 0 (like "payment_actual0"), because they belong to the Application subprocess and are known by the time of decision making. 
As opposed to this, you must not use attributes with suffix > 0 as predictors, because predicting the reopening of a case knowing that there will be additional payments would be trivial.

Likewise, you would also not be allowed to use any derived attributes and events after "inspection" for Q2, because the necessity of an inspection with resulting penalties is just what you want to predict.