The "unusual" status of a sensor means that the currently measured values of the sensor show big differences to the values that are normally retrieved by this sensor. So, for example, if your traffic sensor on a router suddenly shows a huge increase in traffic compared to other days (which might hint on an attack from the outside), PRTG can recognize this and indicates it with the "unusual" status of the sensor.
PRTG calculates the comparison between current and historic monitoring data on a 24 hour basis with internal heuristic methods. See also this article. Please understand that we cannot be more specific on this mathematical approach.
The option "24h average is <80% or >120% of weekday average" means that the current 24 hour average of this sensor is either lower than 80% or higher than 120% than the usually measured values of this sensor.
Consider a traffic sensor that usually measures 100 MB average traffic on a weekday. If you choose this option, it would show an unusual status if the current average is below 80 MB or above 120 MB. So the available options for "Unusual Detection" define the granularity of this sensor status.