top of page

How to deal with outliers?

The Data Q Assistant identifies four different types of outliers:

​

image.png

The sensitivity of extreme values (based on the interquartile range) can be adjusted in the settings menu. Here, also the specific outlier checks can be toggled on/off:

image.png

Dealing with outliers defined as extreme values:

​

  • Always check the value in the context of the other information given in the respective row/column (e.g., for time series check the previous and subsequent values, for a selling price the number cannot be negative, ...)

  • For all numbers that seem to be incorrect, contact the respective expert colleague or include the right number yourself

 

Dealing with outliers defined as values with an unusual sign:

​

  • Always check the value in the context of the other information given in the respective row/column (e.g., for time series check the previous and subsequent values, for a selling price the number cannot be negative, ...)

  • Check if the number 'per se' is correct, but there is just a wrong sign by accident

  • If the number seems to be incorrect independent of the sign, contact the respective expert colleague or include the right number yourself

​

Adjusted numbers will always be highlighted in bright green, so it is easy to trace back all changes.

bottom of page