Mastering Data Scaling: The Only Guide You’ll Ever Need (Straight from My Journey)

Author(s): Suraj Jha

Originally published on Towards AI.

How I Finally Conquered Data Scaling: Learn from My Real-World Experience

This member-only story is on us. Upgrade to access all of Medium.

Scaling is one of the fundamental steps in data preprocessing.

It becomes a must when we are especially working with a dataset that is going to work as an input for machine learning models.

Scaling transforms raw data into a suitable format that machine learning models can understand.

From improving performance to ensuring meaningful comparisons between features, it manages all perfectly.

Let’s witness the magic of scaling in data cleaning and data preprocessing.

It's better to understand it with a real-world example.

Imagine you have a dataset where one of its features is age (ranging from 0 to 60) and another feature is their annual income (ranging from $10,000 to $5,000,000).

ML algorithms like gradient descent struggle with this kind of feature because large-scale features often dominate the optimization process. It ultimately leads to skewed results.

Scaling makes sure that all features are treated equally by ML models, because that way we can enhance the model’s accuracy and convergence speed.

One scaling type doesn’t solve all problems, so we have four types of scaling methods.

It is useful when we know the bounds (min and max values) of the data and want to draw the relationship between… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI