# Is it necessary to deal with the outliers if we perform Normalisation on the data?

I am wondering, if it is necessary to remove outliers from the dataset if we perform Normalisation on the data as after Normalisation, all the values will shrink to value between 0 and 1. So, is it necessary?

Normalisation is used to transform all variables in the data to a same range. It doesn't solve the problem caused by outliers.

Just to give an example,

Suppose, following are the data-points:

0,2,5,10,15,20,22,24,990,1000


It is clear that 990 and 1000 are the outliers here.

Using Min-max Normalisation, these data-points will transform to the following:

0, 0.002, 0.005, 0.01, 0.015, 0.02, 0.022, 0.024, 0.99, 1


So, as you can see the after normalisation also, the outliers remains outliers. Only the range is changed.

Here, I have used Min-max Normalisation, but any normalisation technique will change the range only, and in all the cases outliers will remain outliers after normalisation.

• I don't get it. How did you get negative data points after Normalising the dataset. You should probably get a value between 0 and 1. I didn't say standardisation, which you have put as an example. – Akash Dubey Sep 11 '18 at 9:45
• Sorry, It was a mistake. I've updated my answer. – Sanjay Chandlekar Sep 11 '18 at 9:58