Normalizing numerical data¶

The importance of normalizing data¶

• If your model is based on several numerical attributes - are they comparable?
  • Example: ages mayrange from 0-100 and icomes from 0 to billions
  • Some models may not perform well when different attributes are on very different scales
  • It can result in some attributes counting more than others
  • Bias in the attributes can also be a problem

Examples¶

• Scikit-learn's PCA implementation has a "whiten" option that does tihs for you. use it.
• Scikit-learn has a preprocessing module with hany normalize and scale functions
• Your data may have "yes" and "no" that needs to be converted to "1" and "0".

Read the docks¶

• Most data mining and machine learning techniques work fine with raw, un-normalized data
• But double check the one you're using before you start
• Don't forget to re-scale your results when you're done.