Index Ai Academy
Foundation · Statistics and information

Describing data: center, spread and scale

Mean, median and mode, variance and standard deviation, percentiles, outliers, and standardizing features.

About 42 minutes · Free lesson

What it covers6

Why it matters

Lesson 92 summarized a distribution by its expected value and spread. Real data comes as a column of numbers instead: nine days of tent rentals, ten thousand response times, a million house prices. Nobody reads the whole column, so a few numbers have to stand in for it: where it sits, how widely it spreads, and what its edges look like. Choosing the right few numbers is a skill. One very large value can drag the average far from anything typical, and a feature measured in thousands can drown out one measured in tenths, so a model trained on raw columns may barely learn at all. This lesson builds each summary, shows when each one misleads, and ends with the two jobs it does every day in machine learning: putting features on the same scale, and reporting how fast a system is with percentiles rather than an average.

What you'll be able to do

  • Summarize a column of numbers by its center (mean, median, mode) and its spread (range, variance, standard deviation, percentiles), and read all three off a histogram
  • See why deviations are squared, what an outlier does to each summary, and the difference between dividing by n and by n − 1
  • Compute each summary with a loop, then with np.mean, np.median, np.std and np.percentile, and draw the column with Matplotlib
  • Standardize features so training works, in NumPy and PyTorch, and report response times as percentiles
Quick check
Have a guess before you start. Nine days of tent rentals average 14. On a tenth day a school group rents 60 tents. What happens to the average?
  1. It stays at 14
  2. It jumps well above anything typical
  3. It rises by a tiny amount
  4. It falls
Show the answer

It jumps well above anything typical

One huge value moves the mean a lot: from 14 to 18.6 here, above every one of the ordinary days.

The math: a column in a few numbers

Purpose. Say where a column of numbers sits and how widely it spreads, in a few numbers.

The center. Nine days of rentals: 12, 15, 11, 14, 15, 15, 17, 16, 11. Three numbers describe their middle:

  • The mean adds them and divides by how many: 126 ÷ 9 = 14.
  • The median sorts them (11, 11, 12, 14, 15, 15, 15, 16, 17) and takes the middle one: 15. With an even count, it's the average of the two middle values.
  • The mode is the most common value: 15, which came up three times.

The mean is lesson 92's expected value with every day weighted equally, 1/9 each. A weighted mean gives some values more say than others, and it comes back in lesson 113, where attention averages words by weights.

Quick check
From lesson 92: if each of the nine days gets probability 1/9, what does the expected value Σ xᵢpᵢ become?
  1. The largest value
  2. The median
  3. The sum of the values
  4. The plain mean, sum ÷ 9
Show the answer

The plain mean, sum ÷ 9

Σ xᵢ × 1/9 = (Σ xᵢ)/9: the plain average is an expected value with equal weights.

The spread, built from a failure. The range, largest minus smallest, is 17 − 11 = 6, but it only looks at the two edges. To use every value, measure each one's distance from the mean, its deviation. The obvious next move fails: the deviations always add up to 0, the positives exactly canceling the negatives, so their average is always 0. Squaring them first removes the signs.

The picture. Step through it for the nine days, as boxes of numbers and as a drawing. The drawing is a histogram: one block per day, stacked over the number of tents that day rented, so the three 15s make the tallest pile. The mean appears as the line the blocks balance on, then each block shows its deviation and its square, and the last step shades one standard deviation either side of the mean and adds the median.

Try it now

Then change the 17 in values to 60 and press Use these numbers. The mean line leaves the pile of ordinary days behind, the band grows wide enough to hold every ordinary day, and the median stays put.

Symbols and formula. For n values x₁ to xₙ:

xˉ=1n∑ixis2=1n∑i(xi−xˉ)2z=x−xˉs\bar{x} = \frac{1}{n}\sum_i x_i \qquad s^2 = \frac{1}{n}\sum_i (x_i - \bar{x})^2 \qquad z = \frac{x - \bar{x}}{s}

In plain words: the mean is the balance point, the variance the average squared distance from it, s its square root back in the data's units, and z how many s's a value sits from the mean.

For the nine days: s² = 38 ÷ 9 = 4.22 and s = 2.05 tents. The 17-tent day has z = (17 − 14) ÷ 2.05 = 1.46: about one and a half standard deviations above a typical day. Percentiles split the sorted column: the 50th is the median, and the 95th is the value 95% of the column sits below.

Outliers. Add the 60-tent school day. The mean jumps from 14 to 18.6, above every ordinary day, and s jumps from 2.05 to 13.9. The median barely notices: 15, the same as before. That's why house prices and incomes are reported as medians.

Quick check
What is the median of 3, 9, 4, 7?
  1. 5.5
  2. 4
  3. 5.75
  4. 7
Show the answer

5.5

Sorted: 3, 4, 7, 9. With an even count, average the two middle values: (4 + 7) ÷ 2 = 5.5.

In your own words
In your own words: why does an outlier move the mean so much more than the median?

Write a sentence or two, then compare it with a good answer in the lesson.

In Python: loops, then one call each

Purpose. Compute every summary, first by hand-written loops, then with NumPy's one-liners.

import numpy as np

r = [12, 15, 11, 14, 15, 15, 17, 16, 11]
total = 0
for v in r:
    total = total + v
mean = total / len(r)
squares = 0
for v in r:
    squares = squares + (v - mean) ** 2
print(mean, round((squares / len(r)) ** 0.5, 4))

x = np.array(r, dtype=float)
print(x.mean(), np.median(x), round(float(x.std()), 4), np.percentile(x, 50))
It runs right here in your browser, with nothing to install.

The loop and the one-liners agree: mean 14, standard deviation 2.0548, and the 50th percentile is the median, 15.

Code flow. Step through the two loops one line at a time. The first adds up the days for the mean; the second adds up the squared deviations, the same squares the blocks showed in the math part's picture.

A named mistake: n or n − 1. np.std(x) divides by n. Many other tools, and np.std(x, ddof=1), divide by n − 1 instead, which corrects a small underestimate when the column is a sample from something bigger: the sample's own mean sits a little closer to its values than the true mean would. For the nine days the two give 2.05 and 2.18. Neither is wrong, but mixing them makes hand checks fail, so say which one you used. This module divides by n unless it says otherwise.

Worked example

Worked example: summaries with and without an outlier

Summarize the nine days, add the 60-tent day, and summarize again.

import numpy as np

r = np.array([12, 15, 11, 14, 15, 15, 17, 16, 11], dtype=float)
r2 = np.append(r, 60.0)
print("mean:", r.mean(), r2.mean())
print("median:", np.median(r), np.median(r2))
print("sd:", round(float(r.std()), 2), round(float(r2.std()), 2))
print("sd, n - 1:", round(float(r.std(ddof=1)), 2))
In the lesson you can change this and run it.

One extra day moves the mean by 4.6 tents and multiplies the standard deviation by almost seven, from 2.05 to 13.94, while the median doesn't move at all. Neither summary is wrong; they answer different questions. The mean answers "how many tents in all, shared out per day?", useful for ordering stock. The median answers "what does a typical day look like?", which the school group shouldn't change. The last line shows the n − 1 version for comparison.

Quick check
What does this print?
import numpy as np

x = np.array([2.0, 4.0, 4.0, 4.0, 5.0, 5.0, 7.0, 9.0])
print(x.mean(), x.std(), np.median(x))
  1. 5.0 2.0 4.0
  2. 5.0 4.0 4.5
  3. 4.5 2.0 5.0
  4. 5.0 2.0 4.5
Show the answer

5.0 2.0 4.5

The mean is 40 ÷ 8 = 5; the squared deviations add to 32, so the variance is 4 and s = 2; the median averages the middle pair, 4 and 5.

Draw it. A histogram of the nine days, with the mean and the median marked. plt.hist sorts the values into bins and draws a bar for each; the bin edges at 10.5, 11.5 and so on center each bar on a whole number of tents. plt.axvline draws a vertical line: solid for the mean, dashed for the median.

import numpy as np
import matplotlib.pyplot as plt
r = np.array([12, 15, 11, 14, 15, 15, 17, 16, 11])
plt.hist(r, bins=np.arange(10.5, 18))
plt.axvline(r.mean(), color="black")
plt.axvline(np.median(r), linestyle="--")
plt.xlabel("tents rented")
plt.ylabel("days")
plt.show()
Quick check
Read the plot: the solid line, the mean, sits left of the dashed median. Why?
  1. The mean leaves out the three 15s
  2. The low days sit further below the middle
  3. More of the days rented fewer than 14 tents
  4. The median is always bigger than the mean
Show the answer

The low days sit further below the middle

The mean is the balance point, so distance counts. The 11, 11 and 12 sit 4, 4 and 3 tents below the median, while the high days, 16 and 17, sit only 1 and 2 above it, so the mean is pulled left.

Drill

Standardize a column

Fix standardize(x) so it returns each value's z-score: subtract the mean, then divide by the standard deviation. Then set z for the temperatures, and z_mean and z_sd to z's mean and standard deviation, each rounded to 6 places.

import numpy as np

def standardize(x):
    return x - x.mean()

t = np.array([20.0, 25.0, 30.0, 22.0, 28.0])
z = standardize(t)
z_mean = round(float(z.mean()), 6)
z_sd = round(float(z.std()), 6)
print(z.round(3).tolist(), z_mean, z_sd)
In the lesson, your code is checked as you write it.
In your own words
Why does a standardized column always have mean 0 and standard deviation 1, whatever the original numbers were?

Write a sentence or two, then compare it with a good answer in the lesson.

In AI and ML: scale before training, percentiles for speed

Purpose. Two everyday uses: putting features on the same scale, and reporting how fast a system is.

Scale. Lesson 79's training loop can fit rentals from temperature. Here it runs for 100 steps on five days, first on the raw temperatures, around 25, and then on the same temperatures standardized:

import numpy as np

t = np.array([20.0, 25.0, 30.0, 22.0, 28.0])   # temperature
y = np.array([30.0, 45.0, 60.0, 40.0, 55.0])   # rentals

def train(x, lr):
    w, b = 0.0, 0.0
    for step in range(100):
        err = w * x + b - y
        w = w - lr * 2 * np.mean(err * x)
        b = b - lr * 2 * np.mean(err)
    return float(np.mean((w * x + b - y) ** 2))

z = (t - t.mean()) / t.std()
print(round(train(t, 0.001), 2), train(t, 0.002))
print(round(train(z, 0.1), 2), round(t.mean(), 2), round(t.std(), 2))

On raw temperatures the loss bowl is steep in w and shallow in b, so a learning rate small enough to stay stable in w, 0.001, crawls in b: after 100 steps the loss is still 16.18. Double it to 0.002 and training blows up, to a loss past 10⁴¹. Standardize the temperatures (mean 25, s 3.69) and the bowl turns round: a learning rate of 0.1 reaches 2.16 in the same 100 steps. With many features the effect is stronger, because a feature in the thousands would drown out one in tenths. Real pipelines standardize every column with the training set's own mean and s, and reuse those numbers on new data.

Quick check
Temperature runs from 15 to 35 and income from 20,000 to 90,000. Why standardize both before training?
  1. Standardizing makes the model more accurate on its own
  2. To remove the outliers
  3. To make every value positive
  4. So neither feature dominates just because of its units
Show the answer

So neither feature dominates just because of its units

On raw scales, income's huge numbers set the shape of the loss bowl. After standardizing, both features range over similar numbers, so one learning rate suits both.

Percentiles. Response times are lopsided: most requests are quick and a few are slow. Services report the median and the 95th or 99th percentile, "p95", because users feel the slow tail that an average hides. Here are ten times in milliseconds, one of them a slow 400. Predict their mean, median and p95 before you check.

Quick check
What does this print?
import numpy as np

ms = np.array([120, 95, 110, 400, 105, 98, 130, 102, 115, 99], dtype=float)
print(ms.mean(), np.median(ms), round(float(np.percentile(ms, 95)), 1))
  1. 137.4 107.5 278.5
  2. 137.4 110.0 278.5
  3. 107.5 137.4 278.5
  4. 137.4 107.5 400.0
Show the answer

137.4 107.5 278.5

The one slow request pulls the mean to 137.4; the median is 107.5. The 95th percentile falls between 130 and 400, at 278.5.

Drill

Standardize a table, report a percentile

X has one row per day and one column per feature (temperature, rentals). Set Z to X standardized column by column (use axis=0), ok to whether every column of Z has mean 0 and standard deviation 1 (round both to 6 places), and p95 to the 95th percentile of the response times, rounded to 1 place.

import numpy as np

X = np.array([[20, 30], [25, 45], [30, 60], [22, 40], [28, 55]], dtype=float)
ms = np.array([120, 95, 110, 400, 105, 98, 130, 102, 115, 99], dtype=float)
Z = X
ok = False
p95 = 0
print(Z.round(2).tolist(), ok, p95)
In the lesson, your code is checked as you write it.

In PyTorch. The same table, standardized with tensors: x.mean(dim=0) and x.std(dim=0) give one number per column, as axis=0 does in NumPy. Watch the standard deviations, though. PyTorch's std divides by n − 1 unless you pass correction=0, which divides by n, as NumPy does. torch.quantile gives percentiles, with the percentile written as a share: 0.95 for p95.

import torch

X = torch.tensor([[20.0, 30.0], [25.0, 45.0], [30.0, 60.0], [22.0, 40.0], [28.0, 55.0]])
print(X.std(dim=0))
print(X.std(dim=0, correction=0))
Z = (X - X.mean(dim=0)) / X.std(dim=0, correction=0)
print(Z.round(decimals=2))
ms = torch.tensor([120.0, 95.0, 110.0, 400.0, 105.0, 98.0, 130.0, 102.0, 115.0, 99.0])
print(round(torch.quantile(ms, 0.95).item(), 1), torch.quantile(ms, 0.5).item(), torch.median(ms).item())

With correction=0, Z is exactly the table your drill printed, and p95 is 278.5 again. Two traps show up. The default std, 4.1231 and 11.9373, is the n − 1 kind, so z-scores made with it come out a little smaller than NumPy's. And torch.median gives 105.0, not 107.5: with an even count, PyTorch returns the lower of the two middle values instead of their average. torch.quantile(ms, 0.5) gives 107.5, the median NumPy reports.

In the real tools, scikit-learn does the standardizing as a reusable step. fit learns each column's mean and standard deviation from the training data, and transform applies them to any rows, new days included:

In your own words
A dashboard shows a service's average response time as 137 ms. What would you ask for instead, and why?

Write a sentence or two, then compare it with a good answer in the lesson.

Goes deeper in

Practice what you just read

  • 5 code examples you can edit and run
  • a practice exercise that checks your code
  • a 5-question quiz
  • 9 review cards for spaced repetition

Try a quiz question

House prices in a town are reported as a median, not a mean. Why?

  1. A few costly houses pull the mean above typical
  2. The median is easier to compute
  3. The mean can't be computed for a list of prices
  4. The median is always larger

Answer it at the end of the lesson to see if you're right, and why.

Before this lesson

Keep learning

Where this lesson leads.

575 lessons in thirteen phases. 176 are free without an account, and a free account opens the rest. See every lesson.

Foundation · about 96 hours of study

Foundation math

From whole numbers to the math inside a neural network: each idea by hand, in Python, and in AI 118 lessons All free
  1. Numbers and place value FreeWhat a digit's place is worth, the number line, and Python as a calculator that never tires.36 min
  2. Multiplying and dividing: groups FreeEqual groups, quotient and remainder, and cutting training data into batches.35 min
  3. Order of operations, and names for numbers FreeBrackets, powers, then multiply and divide, then add and subtract; variables hold numbers.36 min
  4. Numbers below zero FreeThe number line through zero, sign rules, distance from zero, and an error's direction and size.37 min
  5. Fractions: equal parts, equal fractions, adding FreePieces of a whole, equal fractions, and re-cutting to add; accuracy as a fraction.45 min
  6. Multiplying and dividing fractions FreeA share of a share, dividing by flipping, and why chances in a row get small.35 min
  7. Decimals and rounding FreeFractions in place value, rounding, and why 0.1 + 0.2 isn't exactly 0.3.36 min
  8. Percentages FreePer hundred, percent change, percentage points, and the accuracy and error rates of models.35 min
  9. Ratios, rates and proportion FreeSharing in a ratio, rates per unit, scaling, and splitting data 80/10/10.35 min
  10. Powers and roots FreeRepeated multiplying, the laws of exponents, square roots, and squared error.43 min
  11. Very big and very small numbers FreePowers of ten, e-notation like 3e-4, orders of magnitude, and a model's size in memory.36 min
  12. Bits, bytes and floats: how numbers are stored FreeBase 2, what n bits can hold, a float as sign, power and digits, and a tensor's memory.35 min
  13. Review: find the slip (numbers) FreeBroken calculations to fix, mixed practice from the whole module, and its formula sheet.34 min
  14. Letters for numbers FreeVariables, terms and like terms, substitution, and a model's formula as a Python function.41 min
  15. Expanding and factoring FreeBrackets as areas, pulling out common factors, and a squared error expanded into a bowl.41 min
  16. Solving equations: keep the balance FreeThe same move on both sides, undoing steps in reverse, and running a model's formula backward.41 min
  17. Rearranging a formula FreeMaking any letter the subject, round-trip checks, and planning a training run.36 min
  18. Inequalities and true-or-false FreeRanges, flipping the sign, if and elif, and the thresholds, clips and ReLU inside a model.42 min
  19. Functions: a rule from input to output Freef(x) notation, tables and graphs, domain and range, loops over lists, and loss as a function.42 min
  20. Chaining and undoing functions Freeg(f(x)), inverses that undo, scaling and unscaling, and a network as a chain.42 min
  21. Straight lines: slope and intercept FreeRise over run, the line through two points, and a model's weight and bias as slope and intercept.42 min
  22. Two equations, two unknowns FreeSubstitution and elimination, where lines cross, and why models are fitted, not solved.42 min
  23. Quadratics: bowls and their lowest point FreeRoots by factoring and by formula, the discriminant, the vertex, and the best weight of a model.42 min
  24. Exponentials: growth and decay FreeThe same factor each step, the number e, and decaying learning rates.43 min
  25. Logarithms: the power question FreeTo what power?, the laws of logs, and adding log-probabilities instead of multiplying.45 min
  26. Sequences and Σ FreeArithmetic and geometric sequences, Σ as a loop that adds, and mean squared error.40 min
  27. Weighted and moving averages FreeWeights that add to 1, a window average, the exponential moving average, and smoothing a loss curve.35 min
  28. Sets, and reading the symbols FreeSets, union and intersection, ℝⁿ and f : ℝⁿ → ℝᵐ, and reading a paper's first equation.41 min
  29. Review: find the slip (algebra) FreeBroken algebra and code to fix, mixed practice from the whole module, and its formula sheet.34 min
  30. Transposing step by step: undo in reverse order FreeA formula's layers, the opposite of each, peeling from the outside in, conditions, and round-trip checks.35 min
  31. Formulas with fractions: clear them first FreeClearing denominators, cross-multiplying, a letter in the bottom, and reciprocal formulas like 1/R = 1/R₁ + 1/R₂.35 min
  32. When the letter appears twice: collect and factor FreeGathering the letter on one side, factoring it out, dividing by the bracket, and rearranging precision and F1.35 min
  33. Squares, roots and powers FreeSquare roots with ±, squaring to undo a root, cube and nth roots, and the √d inside attention.40 min
  34. When the letter is squared and plain: quadratics FreeRearranging into ax² + bx + c = 0, the formula, completing the square, and choosing the root that fits.35 min
  35. Letters in the power: exponentials and logarithms FreeTaking logs to bring a power down, exponentiating to free a letter, and the logit that undoes the sigmoid.40 min
  36. Rearranging inequalities: when the sign flips FreeThe four moves on an inequality, negatives and reciprocals that flip it, logs below 1, and sample sizes.35 min
  37. Inverse functions, lines and scaling FreeUndoing a function by making x the subject, domains that allow it, line forms, and unscaling predictions.40 min
  38. Sequences and sums: find n, a, d and r FreeEvery letter of the arithmetic and geometric formulas as the subject, and an average's window from its decay.40 min
  39. Combining formulas: substitute and eliminate FreePutting one formula inside another to remove a letter, and sizing a training run from a time or money budget.35 min
  40. Review: rearrange anything (algebra) FreeBroken rearrangements to fix, mixed practice from the whole algebra block, a strategy chart, and its formula sheet.35 min
  41. Coordinates and distance FreePoints as (x, y), Pythagoras, distance in any number of coordinates, and nearest neighbors.42 min
  42. Angles, degrees and radians FreeMeasuring a turn, converting units, arcs, and the angle between two arrows.36 min
  43. Sine and cosine FreeAcross and up: right triangles, the unit circle, and directions and rotations.42 min
  44. Waves: height, speed and shift FreeAmplitude, period and phase, sampling a wave, and position codes in transformers.37 min
  45. Review: find the slip (shapes and waves) FreeBroken geometry and code to fix, mixed practice from the module, and its formula sheet.29 min
  46. Angles, triangles and waves: rearranging with sin, cos and tan FreeSides and angles from SOH CAH TOA, principal values, arcs, polar coordinates, and a wave's time and period.45 min
  47. Vectors: lists of numbers that point FreeA list, a point and an arrow; components, dimension, NumPy arrays, and data as vectors.42 min
  48. Adding and scaling vectors FreeComponent-wise sums, scalar multiples, mixes, and the weight update w − ηg.43 min
  49. Length and distance: norms FreeL2, L1 and max norms, unit vectors, distance, and the norms inside losses and clipping.42 min
  50. The dot product FreeMultiply in pairs and add; a·a as squared length; a @ b; one neuron as a dot product.43 min
  51. Angle, cosine similarity and projection FreeDirection without length, perpendicular vectors, shadows, and search by meaning.45 min
  52. Perpendicular directions: bases and coordinates FreeOrthonormal directions, coordinates as dot products, rebuilding a vector, and taking a direction out.35 min
  53. Matrices: tables of numbers FreeRows, columns and entries; two-dimensional arrays and slices; a data set as a matrix.41 min
  54. Adding, scaling and transposing matrices FreeEntry-by-entry sums and scalings, the transpose, mean images, and tables turned around.36 min
  55. A matrix times a vector FreeEach row dotted with the vector, the size rule, A @ x, and a layer acting on one example.42 min
  56. Matrix multiplication FreeRow meets column, the size rule, order matters, and a batch through a layer: X W + b.45 min
  57. Identity, diagonal, triangular and symmetric FreeFour patterns and their jobs: rescaling, distance tables, and the causal mask.37 min
  58. Matrices as movements of space FreeStretches, flips, shears and rotations; columns as landing spots; chaining; a layer moving data.42 min
  59. Determinant and inverse FreeThe area factor ad − bc, singular matrices, undoing a matrix, and redundant features.42 min
  60. Solving systems by elimination FreeAugmented matrices, row moves, back-substitution, the three outcomes, and fitting weights.45 min
  61. Independence, span and rank FreeRedundant vectors, span, rank, outer products, and low-rank updates like LoRA.42 min
  62. Eigenvectors: directions that only stretch FreeAv = λv, checking candidates, np.linalg.eig, and the main direction of a data cloud.41 min
  63. SVD and PCA: the important directions first FreeA matrix as pieces by size, the best rank-k copy, compression, and PCA maps of data.45 min
  64. Tensors, shapes and broadcasting FreeAxes and shapes, reshape, axis=, broadcasting, and shapes through a small model.43 min
  65. One-hot vectors and lookup tables: embeddings FreeA one-hot row picks a table's row, a sentence looked up as vectors, and a score for every word.35 min
  66. Sliding dot products: convolution FreeA small pattern slid along a list and over a grid, output sizes, stride and padding, and shared weights.40 min
  67. Review: find the slip (vectors and matrices) FreeShape slips to fix, mixed practice from the module, and its formula sheet with shapes.36 min
  68. Vector equations: rearranging with arrows FreeSolving for a vector or a scale, recovering a gradient from two weight snapshots, and an angle from a dot product.40 min
  69. Matrix equations: multiply by the inverse, on the correct side FreeNo dividing by a matrix: inverses from the left or right, order, transposes, and the normal equations.40 min
  70. Rates of change FreeChange over change, secants, rates from readings with np.diff, and reading a loss curve.41 min
  71. Limits: what a value heads toward FreeApproaching from both sides, holes, corners and jumps, epsilon, and judging activations.37 min
  72. The derivative: slope at a point FreeShrinking secants, f′(x) and dy/dx, Δy ≈ f′Δx, and which way a weight should move.45 min
  73. Derivative rules FreeConstants, powers, multiples, sums, eˣ and ln x, checked numerically, and a squared error's slope.41 min
  74. The product rule and the chain rule FreeProducts and chains of functions, forward then backward, and one neuron's dL/dw.45 min
  75. Lowest points and curvature Freef′ = 0, the second-derivative test, tables, local and global minima, and convex losses.43 min
  76. Standing in for a curve near a point FreeTangent lines, a curvature term, growing error, and why learning rates must be small.37 min
  77. Partial derivatives FreeOne input at a time, ∂, nudging in Python, and the two slopes of a line's loss.41 min
  78. The gradient FreeAll the slopes as a vector, steepest climb and descent, a numerical gradient, and the loss gradient.42 min
  79. Gradient descent FreeStepping against the gradient, the learning rate, a training loop, and fitting a line.45 min
  80. Faster descent: momentum and Adam FreeMomentum and Adam by hand, the bias correction, and three optimizers down one narrow valley.40 min
  81. The chain rule with many paths: backpropagation by hand FreeGraphs, multiply along paths and add across them, gradient checks, and a 2-2-1 network by hand.45 min
  82. Automatic differentiation: how PyTorch finds every slope FreeA tape of steps walked backward, and requires_grad, backward(), .grad, zero_grad() and no_grad().40 min
  83. Jacobians and Hessians FreeSlopes of vector functions, a layer's Jacobian, second partials, and bowls, domes and saddles.42 min
  84. Integrals: adding up thin slices FreeStrip and trapezoid sums, summing undoes differencing, probability as area, and AUC.43 min
  85. Review: find the slip (calculus) FreeCalculus slips to fix, mixed practice from the module, and its formula sheet.36 min
  86. Rearranging in calculus: slopes, steps and constants FreeSetting a slope to zero, Newton's step, the chain rule turned around, constants from a point, and learning rates.45 min
  87. Counting: choices, orders and groups FreeThe counting principle, factorials, ordered picks and groups, and why attention's cost grows with the square of the length.43 min
  88. Probability: from counts to chances FreeEqually likely outcomes, the complement, distributions, seeded simulation, and a next-word model.43 min
  89. Combining events: and, or, independent FreeThe or rule, independence, probability trees, drawing with and without replacement, and chains of steps.42 min
  90. Conditional probability and Bayes' rule FreeConditionals from tables, why P(A | B) isn't P(B | A), Bayes' rule, and precision and recall.45 min
  91. Random variables and distributions FreeUncertain numbers, the probability of a range, sampling by slices, and sampling versus argmax.42 min
  92. Expected value, variance and standard deviation FreeProbability-weighted averages, spread in the data's own units, and a training loss as an expected value.42 min
  93. Bernoulli, binomial, uniform and normal FreeYes-or-no trials, counts of yeses, the 68-95-99.7 rule, z-scores, and an accuracy's error bar.45 min
  94. Two variables at once: covariance and correlation FreeJoint tables and marginals, covariance, correlation, the covariance matrix, and features that carry the same information.45 min
  95. Review: find the slip (probability) FreeProbability slips to fix, mixed practice from the module, and its formula sheet.36 min
  96. Probability formulas: solve for the missing chance FreeConditionals, Bayes' rule for the prior, the or rule, expected values, variance, and z-scores turned around.45 min
  97. Describing data: center, spread and scale FreeMean, median and mode, variance and standard deviation, percentiles, outliers, and standardizing features.42 min
  98. Samples, the law of large numbers and the central limit theorem FreeSamples and populations, the standard error σ/√n, the bell of averages, and error bars on test scores.43 min
  99. Is the difference real? Comparing two models FreePaired differences, a bootstrap interval, what a p-value is and isn't, and why many tries make lucky wins.40 min
  100. Maximum likelihood FreeLikelihoods as products, log-likelihoods as sums, grid search, underflow, and the negative log-likelihood loss.45 min
  101. Linear regression: the best line FreeResiduals, least squares by formula, three roads to one line, the matrix form, and train versus test error.45 min
  102. Logistic regression: a line that answers yes or no FreeThe sigmoid, the log loss, its slope p − y, a training loop, and choosing a threshold.45 min
  103. Scoring a classifier: confusion table, precision, recall, F1 FreeThe four cells, accuracy, precision, recall and F1, and what moving a threshold does to each.35 min
  104. Softmax: scores into probabilities FreeExponentiate and share out, shifting and temperature, the safe softmax, and a classifier's last step.43 min
  105. Surprise, entropy and cross-entropy FreeSurprise in bits, entropy, cross-entropy and KL divergence, perplexity, and a language model's loss.45 min
  106. KL divergence: how far one distribution is from another FreeKL by hand, cross-entropy as entropy plus KL, why it isn't symmetric, and a soft-target loss.35 min
  107. Review: find the slip (statistics and information) FreeStatistics slips to fix, mixed practice from the module, and its formula sheet.38 min
  108. Statistics and information: n, z, logits and bits FreeSample sizes from error bars, inputs from probabilities, logit gaps from softmax, and chances from bits and losses.40 min
  109. One neuron, start to finish FreeWeighted sum, activation, loss, slopes and update, by hand and in code, and what one straight line can and can't split.45 min
  110. A layer and a small network as matrices FreeTwo layers on a batch, the shape of every matrix, adding a hidden unit, and XOR solved.45 min
  111. Training a small network FreeThe backward pass in matrix form, gradient checks, a full training loop, and reading loss curves.45 min
  112. The same network in PyTorch Freenn.Sequential with the NumPy weights loaded, the same loss and slopes, then SGD and mini-batches.45 min
  113. Attention as matrix math FreeScaled dot-product scores, the causal mask, a row-by-row softmax, and blending value vectors.45 min
  114. Residual connections and layer norm FreeAdding the input back and why its slope keeps learning alive, layer norm by hand, and add and norm.40 min
  115. A transformer block, start to finish FreeA sentence through embeddings, attention, add and norm, a small network and next-word scores, every shape named.45 min
  116. Planning a model: rearranging the formulas of a working network FreeLayer widths from parameter counts, sizes from memory and compute, the attention scale, and a training plan.40 min
  117. Final test: transpose any formula FreeSlips to fix from every topic, mixed practice of every kind of rearranging, the strategy chart and the full sheet.45 min
  118. Capstone: a sign reader from the formulas FreeBuild, train and test a trail-sign classifier with NumPy alone, and report its accuracy with an error bar.45 min
Part of a free course 118 lessons

The Foundation Math For Ai & ML: from arithmetic to attention

Learn the math behind AI from arithmetic up: numbers, algebra, vectors and matrices, calculus, probability and statistics, each shown as numbers moving and checked in Python, ending with a network you build and train from the formulas.

Never trained an AI? Teach a small one to write, right in your browser. It takes about 20 seconds and needs no account. Teach it to write