Features
- Overview
In AI, a "feature" refers to a measurable property or characteristic of data that is used as input for a machine learning (ML) model, essentially acting as a key attribute that the model analyzes to make predictions or classifications; it's a specific piece of information extracted from raw data that helps the AI understand and learn patterns within the data.
AI features highlights:
- Input for models: Features are the inputs that are fed into a ML algorithm to train and make predictions.
- Data transformation: Raw data often needs to be processed and transformed into meaningful features before being used by an AI model.
- Feature engineering: The process of selecting, creating, and modifying features to optimize the performance of a machine learning model is called feature engineering.
- Examples
A feature is an individual measurable property within a recorded dataset. In ML and statistics, features are often called “variables” or “attributes.” Relevant features have a correlation or bearing (called feature importance) on a model's use case.
In a patient medical data set, for example, features can be age, gender, blood pressure, cholesterol levels, and other observed characteristics associated with the patient.
Features can be single variables, derived variables, or combined attributes constructed from underlying data elements. For example, based on measurements of blood pressure, cholesterol levels, and other influencing factors, we can create an "engineered" feature that is categorical and used to identify a group of observations into a risk category for stroke or heart disease.
Other examples,
- In text analysis: Words, phrases, sentiment scores
- In customer data: Age, purchase history, location
- in predicting house price: square footage, number of bedrooms, location, and year built.
- In image recognition: Pixel color values, edge detection, shape characteristics, and texture
[More to come ...]