Introduction

A function is a mathematical relationship that maps each input from a set (called the domain) to exactly one output in another set (called the range). Functions are one of the most fundamental building blocks of mathematics, forming the basis for algebra, calculus, statistics, and virtually every area of machine learning and data science.

From simple linear equations to complex neural network activation functions, understanding functions is essential for modeling relationships between variables and building predictive systems.

Why are Functions Important?

Functions help to:

• Represent relationships between input and output variables
• Model real-world phenomena mathematically
• Form the foundation of calculus, including derivatives and integrals
• Enable machine learning models to map inputs to predictions
• Describe transformations, growth, and decay patterns
• Support algorithm design and computational logic

Identifying a Function

Whiteboard

Loading diagram...

Key Terminology

• Domain: The complete set of possible input values (x) for a function.
• Range: The complete set of possible output values (y) produced by a function.
• Independent Variable: The input variable, typically x.
• Dependent Variable: The output variable, typically y = f(x), which depends on x.
• Mapping: The rule or relationship that connects each input to its corresponding output.

Types of Functions

1. Linear Function

Produces a straight-line graph; has the form f(x) = mx + b.

f(x) = 2x + 3

2. Quadratic Function

Produces a parabolic (U-shaped) curve; has the form f(x) = ax² + bx + c.

f(x) = x² - 4x + 4

3. Exponential Function

Grows or decays rapidly; has the form f(x) = a × bˣ.

f(x) = 2ˣ

4. Logarithmic Function

The inverse of an exponential function; has the form f(x) = logₐ(x).

f(x) = log₂(x)

5. Polynomial Function

Involves terms with non-negative integer exponents of x.

f(x) = 3x³ - 2x² + x - 5

6. Trigonometric Functions

Relate angles to ratios of sides in triangles (e.g., sine, cosine, tangent).

f(x) = sin(x)

7. Piecewise Function

Defined by different expressions depending on the value of x.

f(x) = { x²,  if x < 0
       { 2x,  if x ≥ 0

Function Notation and Evaluation

Example: For f(x) = 3x + 5, find f(4)

f(4) = 3(4) + 5 = 12 + 5 = 17

One-to-One vs Many-to-One Functions

Key Properties of Functions

• Every input in the domain must map to exactly one output.
• A function may be increasing, decreasing, or constant over different intervals.
• Functions can be combined through addition, subtraction, multiplication, division, or composition.
• The inverse of a function reverses the mapping, swapping domain and range.
• Not every curve represents a function — use the vertical line test to check.

Functions in Machine Learning

Where are Functions Used?

Advantages

• Provides a clear, structured way to represent relationships between variables
• Forms the mathematical foundation for calculus and optimization
• Enables predictive modeling in machine learning and statistics
• Supports reusable, modular logic in programming and engineering
• Allows complex systems to be broken down into manageable relationships

Limitations

• Not all real-world relationships can be perfectly modeled by simple functions
• Some functions can be difficult to differentiate or integrate analytically
• Choosing the wrong function type can lead to poor model fit
• Piecewise or discontinuous functions can complicate analysis
• High-degree polynomial functions can overfit data

Real-World Examples

Best Practices

• Always verify a relation is truly a function using the vertical line test.
• Clearly define the domain and range before analyzing a function's behavior.
• Choose the function type that best matches the underlying real-world relationship.
• Simplify or decompose complex functions when possible for easier analysis.
• Understand a function's derivative behavior when using it in optimization contexts.

Interview Tip

A common interview question is:

"What is a function in mathematics, and how does it apply to machine learning?"

A strong answer is:

A function is a relationship that maps each input from a domain to exactly one output in a range, commonly written as y = f(x). In machine learning, functions are everywhere — from the linear function underlying linear regression, to activation functions like sigmoid, ReLU, and softmax that introduce non-linearity in neural networks, to the loss functions used to measure prediction error during training.

Mentioning specific machine learning function types like activation and loss functions makes your answer stronger.

Conclusion

Functions are a foundational mathematical concept that describe how one quantity depends on another, forming the backbone of algebra, calculus, and virtually all machine learning models. Understanding different types of functions and their properties is essential for building, analyzing, and optimizing mathematical and computational systems.