WhatsApp us
IB DP Maths17 Nov 2026

IB Maths AI HL Poisson Distribution: When to Use It, Rescaling the Rate, Sums and Worked Examples

The Poisson distribution models random events at a constant average rate, and it is an AI HL topic in the current course. How to check a Poisson model is reasonable, rescale the mean to a new interval, add independent Poisson variables and solve for the mean, with four worked examples checked to 3 significant figures.

IB Maths AI HL Poisson Distribution: When to Use It, Rescaling the Rate, Sums and Worked Examples

Use the Poisson distribution when you are counting how many times something happens in a fixed stretch of time or space, the events happen one at a time and independently, and the average rate stays constant. You write X ~ Po(m), where m is the mean number of events in that interval. Then P(X = x) = e^(−m) × m^x ÷ x! for x = 0, 1, 2, …, and both the mean and the variance equal m.

In the current IB courses (first assessment 2021), Poisson sits in the additional higher level content of AI HL only, inside Topic 4, statistics and probability. AA students do not meet it. Every AI paper requires a GDC, so the arithmetic is rarely the hard part. The marks go to choosing the right model, the right interval and the right inequality. This guide covers those decisions with four original examples. If you want regular help across the whole course, our one-to-one IB Maths AI HL tutoring starts with a free diagnostic demo.

Is Poisson still in AI HL?

Yes, for anyone sitting exams in 2027 or 2028. The IB says the final assessment of the current AA and AI courses is November 2028. A new curriculum is taught from August 2027, with first exams in May 2029, and the IB lists the Poisson distribution among the content removed from AI HL in that new course. So a student who starts DP in August 2027 will not be examined on Poisson. Anyone on the current course still needs everything below. Our summary of what changes in the 2029 IB maths courses has the full list.

The four conditions, in plain words

Examiners like to ask you to "state an assumption" or "explain why a Poisson model may not be suitable". Have these ready:

  1. Events occur singly. Two events cannot happen at exactly the same instant.
  2. Events are independent. One arrival does not make the next more or less likely.
  3. The rate is constant. The average number per unit time or space does not change across the interval.
  4. You are counting in a fixed interval. There is no fixed number of trials, which is what separates Poisson from binomial.

A good "not suitable" answer is specific to the context. Customers arriving at a café are unlikely to arrive at a constant rate across a whole day, because lunchtime is busier. Cars on a road may arrive in bunches after a traffic signal, which breaks independence.

A quick data check. For a Poisson variable the mean equals the variance. If a frequency table gives a sample mean and variance that are close, a Poisson model is plausible. Suppose a student counts cars passing a gate in 100 separate one-minute intervals:

Cars per minute012345
Frequency2233251352

The mean is 152 ÷ 100 = 1.52 and the variance is about 1.49. These are close, so Po(1.52) is a sensible model. Expected frequencies from Po(1.52) are roughly 21.9, 33.2, 25.3, 12.8, 4.9 and 1.5, which match the data well. If the variance had been much larger than the mean, the counts would be more spread out than Poisson allows.

Worked example 1: the basic probabilities

An online bookshop receives orders at an average rate of 3.5 per hour. Let X be the number of orders in a randomly chosen hour, so X ~ Po(3.5).

(a) P(X = 2).

e^(−3.5) × 3.5² ÷ 2! = 0.185 (3 s.f.). On the GDC this is the Poisson pdf with m = 3.5 and x = 2.

(b) P(X ≤ 4).

Poisson cdf from 0 to 4 = 0.725.

(c) P(X ≥ 6).

The GDC cdf works with "at most", so rewrite: P(X ≥ 6) = 1 − P(X ≤ 5) = 0.142.

(d) P(fewer than 3 orders).

"Fewer than 3" means 0, 1 or 2, so P(X ≤ 2) = 0.321.

Writing the inequality conversion as a line of working earns method marks even if the final keystroke goes wrong.

Worked example 2: rescaling the rate and adding variables

The same shop also receives phone calls at an average of 1.1 per hour, independently of the 2.4 emails it receives per hour.

(a) Find the probability of no orders in a 20-minute slot (orders still at 3.5 per hour).

Rescale the mean to the new interval: 20 minutes is one third of an hour, so m = 3.5 ÷ 3 ≈ 1.167.

P(X = 0) = e^(−1.167) = 0.311.

(b) Find the probability of at least 2 orders in that 20-minute slot.

1 − P(X ≤ 1) = 0.325.

(c) Let T be the total number of emails and calls in a 2-hour period. Find P(T > 9).

The sum of independent Poisson variables is Poisson, with the means added.

Emails in 2 hours: Po(4.8). Calls in 2 hours: Po(2.2). So T ~ Po(7).

P(T > 9) = 1 − P(T ≤ 9) = 0.170.

(d) Find the probability of no calls at all in those 2 hours.

P(C = 0) = e^(−2.2) = 0.111.

The two habits that matter here: rescale m before you calculate anything, and state "independent" when you add the means. Without independence the sum rule does not apply.

Worked example 3: finding the mean from a probability

The number of faults in a 50-metre roll of fabric follows a Poisson distribution. The probability that a roll has no faults is 0.15.

(a) Find the mean number of faults per roll.

P(X = 0) = e^(−m) = 0.15, so m = −ln 0.15 = 1.90 (3 s.f.). Keep the full value, 1.8971…, in your calculator.

(b) Find the probability that a roll has at least 2 faults.

1 − P(X ≤ 1) = 0.565.

(c) A roll is known to have at least one fault. Find the probability it has at least 2.

This is conditional probability:

P(X ≥ 2 | X ≥ 1) = P(X ≥ 2) ÷ P(X ≥ 1) = 0.5654 ÷ 0.85 = 0.665.

When P(X = 0) is given, you can solve with logs by hand. When the given value is something like P(X ≤ 2) = 0.4, there is no neat algebra. Graph the cdf as a function of m on the GDC and find where it meets 0.4.

Worked example 4: Poisson inside a binomial

Typing errors in a report occur at random at an average of 0.8 per page. The report has 10 pages.

(a) Find the probability that exactly 3 pages have no errors.

Step 1, one page: P(no errors on a page) = e^(−0.8) = 0.4493.

Step 2, the pages: let Y be the number of error-free pages, so Y ~ B(10, 0.4493).

P(Y = 3) = 0.167.

(b) Find the probability that the whole report has at most 5 errors.

Total errors in 10 pages ~ Po(8), so P(at most 5) = 0.191.

This two-layer structure, with Poisson for each unit and binomial across the units, is a favourite in longer questions. Underline what is being counted in each part: errors, or pages.

Mistakes our tutors see most often

  • Forgetting to rescale. A rate "per hour" used for a 20-minute question gives a wrong answer that looks reasonable.
  • Inequality slips. "More than 4" is 1 − P(X ≤ 4). "At least 4" is 1 − P(X ≤ 3).
  • Mixing up pdf and cdf on the calculator. The pdf gives one value. The cdf gives a running total from 0.
  • Using Poisson when there is a fixed number of trials. "20 seeds planted" is binomial.
  • Adding means of dependent variables, or not stating the independence assumption.
  • Rounding m early in a find-the-mean question, then losing the accuracy mark later.

Exam technique

  • Define the variable in words and write the distribution, for example "X = number of calls in 2 hours, X ~ Po(2.2)". This line often carries a mark on its own.
  • Write the probability statement before the GDC answer, such as "P(T > 9) = 1 − P(T ≤ 9)".
  • Give answers to 3 significant figures unless told otherwise, and keep unrounded values stored.
  • Context questions want context answers: "the model may not hold because orders are more frequent in the evening" beats "the rate may not be constant".
  • On Paper 3, expect a long problem that builds a model step by step. Poisson can appear as one stage of a bigger statistics story.

Poisson also underpins hypothesis tests for a rate, with critical regions and error probabilities. These are covered in our guide to Poisson hypothesis tests, critical regions and Type II errors, so we do not repeat them here. For deciding between models before you start, see our post on choosing between binomial, normal and Poisson models.

Self-check

  1. X ~ Po(2.6). Find P(X = 3).
  2. Events happen at 4 per day. Find the probability of none in a 6-hour period.
  3. X ~ Po(1.2) and Y ~ Po(2.3) are independent. Find P(X + Y ≥ 4).
  4. X ~ Po(m) and P(X = 1) = P(X = 2). Find m, then P(X = 0).

Answers: 1) 0.218. 2) m = 1, so e^(−1) = 0.368. 3) X + Y ~ Po(3.5), so 1 − P(≤ 3) = 0.463. 4) m e^(−m) = m² e^(−m) ÷ 2 gives m = 2, so P(X = 0) = e^(−2) = 0.135.

Where this leads

Within AI HL, Poisson connects to the rest of the HL statistics: sums and linear combinations of random variables, then confidence intervals and tests. Students who enjoy modelling often find the AI HL calculus beyond the SL basics a natural next step, and the matrix work in our sibling guide to matrix transformations for AI HL uses the same careful set-up habits. For the whole picture, see the full AI HL topic list.

How Maths Bodhi helps

In the free demo, our tutor asks the student to set up two or three Poisson problems aloud. This shows quickly whether the gap is in modelling, inequalities or calculator use. Tutors are matched to the AI HL course, and practice is past-paper style with written working checked every session. Families who want help across all of DP maths can see our IB Diploma maths support.

Home tuition is available in Gurgaon (Gurugram). Students elsewhere in India and abroad learn with us online. You can book a free demo class.

Written and checked by the Maths Bodhi team. Facts checked against official IB sources, November 2026.

Frequently asked questions

Is the Poisson distribution in IB maths AI HL?
Yes, in the current course it is part of the AI HL additional content, and AA students do not study it. The IB lists Poisson among content removed from AI HL in the new course first examined in May 2029, so students sitting exams in 2027 or 2028 still need it.
How do you know when to use a Poisson distribution?
Use it when you are counting random events in a fixed interval of time or space, the events happen singly and independently, and the average rate is constant. If there is a fixed number of trials, use the binomial instead.
Why are the mean and variance equal in a Poisson distribution?
It is a property of the distribution: if X ~ Po(m), both E(X) and Var(X) equal m. In data questions, a sample mean and variance that are close is evidence that a Poisson model is reasonable.
How do I change the Poisson mean for a different time period?
Scale it in proportion to the interval, provided the rate is constant. For example, 3.5 per hour becomes 3.5 ÷ 3 for 20 minutes and 7 for 2 hours.
Can you add two Poisson distributions?
Yes, if they are independent. The sum is Poisson with mean equal to the sum of the means, so Po(4.8) plus Po(2.2) gives Po(7).
Do I need to use the Poisson formula or can I use the GDC?
Every AI paper requires a GDC, so you can use its Poisson pdf and cdf functions. You still need the formula to solve problems such as e^(−m) = 0.15 and to show the probability statement you are calculating.
IB maths AI HL PoissonPoisson distribution IBIB maths AI HL statisticsdiscrete probability IBIB maths tutor Gurgaon

Related guides

Keep exploring this maths topic

IB DP Maths10 Nov 2026

IB Maths AI HL Hypothesis Tests and Confidence Intervals: Poisson, Unbiased Estimates, Type I and II Errors and GDC Workflow

AI HL statistics goes well beyond the SL chi-squared and t-tests. How to use the Poisson distribution, find unbiased estimates, build a confidence interval for a mean, run binomial, Poisson, normal and correlation tests, and calculate Type I and Type II error probabilities, with worked examples.

IB maths AI HL statisticsIB maths hypothesis testingconfidence intervals IBPoisson distribution IB

By Maths Bodhi Team

Read guide

IB DP Maths17 Nov 2026

IB Maths AI HL Topics and Syllabus: The Course Grouped Into Four Modelling Strands, With Hours, Papers and a Checklist

The IB Maths Applications and Interpretation HL syllabus grouped into four modelling strands (change over time, data and inference, networks and states, space and motion), with the IB's teaching hours, the three papers, a tick-off checklist and the changes for the course first examined in 2029.

IB maths AI HL topicsIB maths AI HL syllabusAI HL modellingAI HL checklist

By Maths Bodhi Team

Read guide

IB DP Maths10 Jul 2026

IB Maths AI HL vs AI SL: Topics, Papers, Workload and How to Choose

IB Maths Applications and Interpretation at HL and SL compared: teaching hours by topic, the HL-only content, how the papers differ, whether AI SL has calculus, who should take which level, and what changes for students starting DP in August 2027.

IB maths AI HLIB maths AI SLAI HL vs AI SLIB maths applications and interpretation

By Maths Bodhi Team

Read guide

IB DP Maths17 Nov 2026

Is IB Maths AI HL Hard? Five Common Beliefs Checked Against the IB's Own Figures

AI HL has a reputation as the easier HL maths. This guide checks that against the IB's published hours and papers, names the topics that catch students out, separates fixable problems from signs of the wrong course, and explains how to switch and what a first month with a tutor involves.

is IB maths AI HL hardIB maths AI HL difficultyAI HL vs AA HLIB maths AI HL workload

By Maths Bodhi Team

Read guide

IB DP Maths10 Nov 2026

IB Maths AI HL Logistic Models and Linearising Data: Semi-Log and Log-Log Graphs, Non-Linear Regression and R²

How IB Maths Applications and Interpretation HL students should handle logistic models, linearise data with semi-log and log-log graphs, judge non-linear regression with R² and residuals, and choose between models, with four worked examples.

IB maths AI HL logistic modellinearising datalog-log graphsnon-linear regression

By Maths Bodhi Team

Read guide

IB DP Maths17 Nov 2026

IB Maths AI HL Matrix Transformations: Rotations, Reflections, Stretches, Enlargements, Composition and the Determinant

In AI HL, a 2 × 2 matrix moves every point of the plane at once. How to build rotation, reflection, stretch and enlargement matrices from scratch, compose them in the right order, read the determinant as an area scale factor and undo a transformation with the inverse, with four checked examples.

IB maths AI HL matricesmatrix transformations IBdeterminant area scale factorAI HL geometry

By Maths Bodhi Team

Read guide