I have a weekly dataset and I have to normalize this data. Data is something like this :

1. week   50
2. week   51
3. week   50
4. week   54
5. week   150
6. week   155
7. week   ...

The important thing is, the difference between week 3 and week 4 (50-54) is not same with week 5 and week 6. And also there is a huge different between week 4 and week 5.

My question is how can i handle all of this things ?

Is the standard normalization functions(for example scikit normalization) can do it for me and should I normalize this data 0-1 or -1 to 1 ?

Sklearn normalization page

NOTE I am working with python and generally scikit-learn library.

Any help is appreciated.

  • 1
    $\begingroup$ What range you normalise to would depend on what you wish to do with the transformed data. $\endgroup$ Commented Feb 2, 2015 at 11:50

1 Answer 1


I would find the unit variance of the all the weeks and then divide by that. Scikit can do this for you using scale.

  • $\begingroup$ Could you please explain more detaily. I will give the list of my data(an array) to scale() function and it will return an array. Then should I divide my real list of data to the other array(which is the result of scale) ? $\endgroup$
    – Batuhan B
    Commented Jan 29, 2015 at 20:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.