202404292158
Status: #idea
Tags: Machine Learning, Deep Learning

Feature Engineering

Features are any measurement on our data that is used for training a model, it could be anything from the price of an item to the time where the item was bought.

Feature Engineering consists in the art of engineering the features that lie dormant in raw data to make them suitable for a model. This is what will happen before we actually train the model.

It consists of finding potential errors in the dataset, filling in a suitable way the empty rows that are bound to be in the dataset, scaling the data in a way that makes sense (potentially normalizing or standardizing the data), removing features that are not useful or even potentially creating new ones out of the existing data.

The goal is to turn the data that is available into its most optimal version so that it can be used in models.

Relevant Links

https://builtin.com/articles/feature-engineering
https://aws.amazon.com/what-is/feature-engineering/