Lesson 9 of 12 · Reading Maths
Summation Notation
Learn to read summation notation: the Σ symbol, the index and its limits, the mean and mean squared error formulas, and the product symbol Π, with each one turned into a Python loop.
- 10 min read
- Beginner
- Reviewed
- How to read the Σ sum symbol and its parts
- How every sum turns into a simple Python loop
- How to read the formulas for the mean and the mean squared error
- What the product symbol Π means
A short way to say "add them all up"
AI papers are full of formulas that add up many values. Writing x₁ + x₂ + x₃ + ... + x₁₀₀₀ would be impossibly long, so mathematicians use one symbol.
n
Σ xᵢ = x₁ + x₂ + x₃ + ... + xₙ
i=1Read it out loud as: "the sum of x-i, for i from 1 to n".
The parts of a sum
Every sum is a for loop
This is the most useful idea in this lesson: Σ is a for loop with a running total.
x = [4, 7, 1, 8] # x₁ to x₄ (Python counts from 0)
total = 0
for i in range(len(x)):
total += x[i]
print("Sum by loop:", total)
print("Sum by sum():", sum(x))The sum can also contain a calculation. For example, the sum of the squares of the numbers 1 to 5:
5
Σ i² = 1 + 4 + 9 + 16 + 25 = 55
i=1total = 0
for i in range(1, 6): # i from 1 to 5: range stops BEFORE 6
total += i ** 2
print(total)Remember that the upper limit is included in maths, but range() stops before its end value. So "from 1 to 5" is range(1, 6).
The mean: your first real formula
1 n
x̄ = - Σ xᵢ
n i=1Read it as: "x-bar equals one over n, times the sum of x-i". In plain words: add them all up, then divide by n.
scores = [72, 85, 90, 64, 89]
n = len(scores)
total = 0
for i in range(n):
total += scores[i]
mean = total / n
print("Mean score:", mean)Mean squared error
Here is a formula you will meet in your very first machine learning model. It measures how far a model's predictions ŷᵢ (y-hat) are from the true values yᵢ.
1 n
MSE = - Σ (yᵢ - ŷᵢ)²
n i=1Read it piece by piece, from the inside out:
yᵢ - ŷᵢ: the error for one example.( )²: square it, so negative and positive errors both count, and big errors count extra.Σ: add up the squared errors for all examples.1/n: divide by the number of examples to get the average.
actual = [3.0, 5.0, 2.5, 7.0]
predicted = [2.5, 5.0, 4.0, 8.0]
n = len(actual)
total = 0
for i in range(n):
error = actual[i] - predicted[i]
total += error ** 2
mse = total / n
print("MSE:", mse)You just turned a formula from a machine learning textbook into working code. That skill, reading a formula and translating it, is what this module is about.
The product symbol
4
Π i = 1 × 2 × 3 × 4 = 24
i=1product = 1 # start at 1 for a product, not 0
for i in range(1, 5):
product *= i
print(product)Products of probabilities appear in AI, and as the Exponents lesson showed, they are usually turned into sums of logarithms.
Common mistakes
Turn a real machine learning formula into a few lines of Python.
# Translate the formula for mean absolute error (MAE) into Python: # # 1 n # MAE = - Σ |yᵢ - ŷᵢ| # n i=1 # # |...| means absolute value: use abs(). actual = [10, 12, 8, 15] predicted = [11, 10, 8, 18] # Output must be exactly: # Total absolute error: 6 # MAE: 1.5
What is the value of Σ (2i) for i from 1 to 3?
Summary
- Σ means add up. The index runs from the lower limit to the upper limit, which is included.
- A subscript such as
xᵢpicks item i from a list, likex[i]in Python. - Every Σ is a
forloop with a running total. - The mean is
(1/n) Σ xᵢ, and the mean squared error is(1/n) Σ (yᵢ - ŷᵢ)². - Π means multiply them all together. Start a running product at 1.
Next: the Greek letters you will meet in AI, and what each one usually means.
Finished reading? Tick it off to track your progress.