17. Video: Feature Scaling Example

Feature Scaling

Here you saw one example where k = 2 for the same dataset. However, depending on the scaling of the features in the dataset, we ended up with 2 completely different clusters.

In most cases, you will want to use Standardized scaling. On the next page, you will get some practice using python to scale data using the two scaling methods we have looked at:

1. Normalizing
2. Standardizing

Standardizing your data (both features in the above case), would assure that we get consistent clustering of our data each time. Manipulating the clusters created by adjusting the feature scalings for either feature is cheating (or likely to provide misleading results)!