Velamuvi AIAcademy

Lesson 7 of 12 · Functions and Graphs

Linear Functions

Understand linear functions y = mx + b: slope as rise over run, the y-intercept, finding a line through two points, and how this is exactly linear regression's model. Interactive graphs and Python.

  • 8 min read
  • Beginner
  • Reviewed
In this lesson you will learn
  • What a linear function is
  • What slope and intercept mean, and how to read them from a graph
  • How to work out the slope from two points
  • How to find the equation of a line through two points
  • Why the straight line is the first model every data scientist learns

Straight lines

The two numbers in y = mx + b control everything about the line:

Move the sliders and watch what each number does:

y = 1x + 1
-4-2024-6-4-20246
y = mx + b
m tilts the line. b slides it up and down.

What you should notice:

  • A positive slope goes up from left to right. A negative slope goes down.
  • A bigger slope (ignoring the sign) means a steeper line.
  • A slope of 0 gives a flat, horizontal line: y never changes.
  • Changing b moves the whole line up or down without tilting it.

Slope as rise over run

            change in y     y₂ - y₁
slope m  =  -----------  =  -------
            change in x     x₂ - x₁

For the points (1, 3) and (4, 9): the rise is 9 - 3 = 6, the run is 4 - 1 = 3, so the slope is 6 ÷ 3 = 2. For every 1 step to the right, the line goes up 2.

-101234560246810(1, 3)(4, 9)
y = 2x + 1
From (1, 3) to (4, 9): run of 3, rise of 6, so the slope is 2.
Try it yourselfPython
def slope(x1, y1, x2, y2):
    return (y2 - y1) / (x2 - x1)


print(slope(1, 3, 4, 9))
print(slope(0, 5, 2, 1))      # going down: negative slope
print(slope(-1, 4, 3, 4))     # flat: slope 0

The line through two points

Once you know the slope, you can find b by putting one point into y = mx + b and solving:

y = mx + b
3 = 2 × 1 + b       use the point (1, 3) and m = 2
b = 1

So the line is y = 2x + 1.

Try it yourselfPython
def line_through(x1, y1, x2, y2):
    m = (y2 - y1) / (x2 - x1)
    b = y1 - m * x1
    return m, b


m, b = line_through(1, 3, 4, 9)
print(f"y = {m}x + {b}")

# Check: does the line really pass through both points?
print(m * 1 + b, m * 4 + b)

A line is a model

This is the first model in machine learning: linear regression. It assumes the output changes at a steady rate with the input.

Try it yourselfPython
# Ice cream sales against temperature, roughly following a line
temperatures = [20, 24, 28, 32]
sales = [110, 150, 190, 230]

m = (sales[1] - sales[0]) / (temperatures[1] - temperatures[0])
b = sales[0] - m * temperatures[0]
print(f"Model: sales = {m} x temperature - {-b}")

for t in [26, 35]:
    print(f"At {t} °C, predicted sales: {m * t + b:.0f}")

The slope has a real meaning: every extra degree sells about 10 more ice creams. Real data never sits perfectly on a line, so linear regression finds the line that is closest to all the points at once. You will learn how in the machine learning courses.

Common mistakes

Practice exercise

Find a line from two points, then use it as a prediction model.

Your solutionPython
# A taxi company charges a fixed amount plus a price per km.
# A 2 km trip costs 74 and a 6 km trip costs 122.
# 1. Find the slope (price per km) and the intercept (fixed charge).
# 2. Predict the cost of a 10 km trip.

# Output must be exactly:
# Price per km: 12.0
# Fixed charge: 50.0
# 10 km costs: 170.0
Knowledge check

In y = -2x + 5, what are the slope and the y-intercept?

Summary

  • A linear function y = mx + b has a straight-line graph.
  • The slope m is how much y changes for each step in x. The y-intercept b is where it crosses the y-axis.
  • Work out the slope with rise over run: (y₂ - y₁) / (x₂ - x₁).
  • Two points are enough to find the whole line.
  • Linear regression fits the best line to data. Its slope and intercept are called the weight and the bias.

Next: curves, and the special functions that let neural networks learn.

Finished reading? Tick it off to track your progress.