Data Preprocessing: 4- Feature Scaling

Feature scaling

Machine learning algorithms usually work with numbers with identical scales. If numbers have different scales, the algorithm may consider those with higher scales to be more important.

Even though all our data is numerical, it is not yet uniformly scaled. For example, the values of most of the columns range between 0 and 3. But “Age” and “Fare” have far bigger scales.

The max method returns the maximum value in a column. As we can see, the oldest passenger was 80 years old, and the highest fare was about 512.