Continuous Random Variables

MIT 6.041 Probabilistic Systems Analysis and Applied Probability, lecture 8, explained in plain English

Want this lecture to stick? In LessonBound you get checked practice questions for every part, can ask about anything you don't get, and can listen with your screen off. Free, no ads.

Learn this lecture free

Watch or read the original lecture: MIT OpenCourseWare.

Part 1 · 0:00–10:00

This part introduces continuous random variables, which can take any real value, and explains how probability is spread out along the number line using density functions.

Key ideas

A continuous random variable can take any real number, not just a list of separate values. Its probability is spread smoothly along the number line. 1:50
Instead of placing chunks of probability at points, we describe a continuous spread with a density function. The density shows how thickly the probability sits at each place. 4:25
To find the probability of an interval, you find the area under the density curve above that interval. In calculus terms, this area is found with an integral. 5:52
A single point has probability 0, because the area over a single point is 0. So the density value at a point does not give that point's probability. 8:30
A valid density can never be negative, and the total area under it over the whole number line must equal 1. 9:13

Words to know

continuous random variable: A number that comes from a random experiment and can take any value in a range, like 2.3751 or 7.0, not just counted steps like 1, 2, 3.
probability density function: A curve that shows how thickly probability is spread along the number line. The area under the curve over an interval gives the probability of that interval.
integral: A calculus tool for adding up a continuously changing amount. For a density, it gives the area under the curve over an interval.
probability mass function: The rule for discrete variables that gives the probability of each exact value. It is the discrete version of a density, where you add instead of integrate.

Example

Imagine spreading one cup of sand over a long table. Some spots get a thick layer and some get only a thin dusting, and the total amount of sand is always one cup. The thickness at each spot works like a density, and the amount of sand over a stretch of table works like a probability.

3 checked practice questions for this part are in the app.

Part 2 · 10:00–20:00

This part explains how probability density works for continuous values, and how to find expected value and variance for them.

Key ideas

A density tells you the probability of a small interval. For a small range of width delta, the probability is about the density times delta, because the density barely changes over that small range. 10:00
A density is not a probability. It is a rate: probability per unit length. So it can be bigger than 1, as long as the total area under the curve is 1. 11:30
Expected value for a continuous variable uses the same idea as before. You swap sums for integrals and probability mass functions for density functions. It still acts like the center of gravity of the density picture. 14:07
The uniform random variable is the simplest continuous example. Every interval of the same length has the same probability, so its height is 1 divided by (b minus a), which makes the total area equal 1. 17:22
For a uniform variable, the expected value is the midpoint of the range, because the picture is symmetric. The variance is found with an integral and comes out to (b minus a) squared, divided by 12. 19:20

Words to know

probability density function: A curve for a continuous variable. The area under it over a range gives the probability of landing in that range.
integral: A way to add up many tiny pieces. For a density, it finds the area under the curve over a range.
expectation: The long-run average value you would get if you repeated an experiment many times. It is also the center of gravity of the density.
uniform random variable: A continuous variable where every interval of the same length is equally likely, so its density is flat over one range.

Example

Imagine a bus that is equally likely to show up at any moment between 10:00 and 10:20. Any 2-minute window has the same chance of containing the bus, so the bus arrival time is uniform. The flat height of the density is 1 divided by 20 minutes, so the total area adds up to 1.

3 checked practice questions for this part are in the app.

Part 3 · 20:00–30:00

This part explains the cumulative distribution function, a single tool that describes both discrete and continuous random variables, and how it connects to density and to mixed cases.

Key ideas

Discrete and continuous random variables use different formulas, with sums and integrals. The cumulative distribution function (CDF) gives one shared way to describe both. 20:44
The CDF at a point x is the probability that the random variable is less than or equal to x. For continuous variables you find it by integrating, and for discrete variables you add up probabilities from the left. 21:05
For a continuous variable, the CDF rises smoothly from 0 to 1. Its derivative gives the density, and the CDF has no jumps because there are no single points with positive probability. 23:39
For a discrete variable, the CDF rises in steps. Each jump is the probability of one exact value, and at a jump point the CDF includes that value's probability. 25:35
A mixed random variable has some probability sitting on single points and some spread over a range. It is neither purely discrete nor purely continuous. 27:46

Words to know

cumulative distribution function (CDF): A rule that tells you the chance a random result is less than or equal to some number. It works for both kinds of random variables.
density: For a continuous variable, a function whose area over a range gives the chance of landing in that range.
point mass: A chunk of probability sitting on one exact value, like a half-pound weight placed on a single point.
mixed random variable: A random variable that has some chance on exact single values and also spreads chance over a continuous range.

Example

Think of a jar of coins where each coin's value is added to a running total as you pick them out. The running total shows how much of the total you have collected so far, just as the CDF shows how much probability has been collected up to a point. A step up each time you get a specific coin matches a discrete CDF, while a smooth climb matches a continuous one.

3 checked practice questions for this part are in the app.

Part 4 · 30:00–40:00

This part introduces the normal (bell-shaped) distribution, explains how its shape and width are set, and shows how averages and spreads change when you multiply or shift a random variable.

Key ideas

The CDF always exists, even for mixed random variables. It rises smoothly where there is a density, and it jumps where a positive amount of probability sits at a single point. 31:42
The normal distribution is very important because when many small independent random pieces are added together, their sum tends to look normal, no matter what the pieces look like. 32:50
The normal density shape comes from the negative exponential of x squared over 2. The curve is highest at the center and its tails fall off quickly. 34:37
The mean mu moves the bell left or right, and sigma controls its width. A smaller sigma gives a narrower bell, and the standard deviation equals sigma, so the variance is sigma squared. 37:55
For any random variable, if you form a linear function Y = aX + b, the expected value is a times the original expected value plus b, and the variance is a squared times the original variance. The constant b does not change the variance. 39:46

Words to know

density: A curve that shows how likely values are in each spot. The probability over a range is the area under the curve in that range.
CDF: Short for cumulative distribution function. It tells you the chance that the random variable is at or below a given value, and it always goes up from 0 to 1.
variance: A number that measures how spread out a random variable's values are around its average.
standard deviation: The typical distance of values from the average. It is the square root of the variance, and for a normal distribution it is the sigma that sets the width.

Example

Think of a bathroom scale reading that is affected by many tiny, unrelated things, like small shifts in your posture, the floor, and the air. Each small effect is random, and when they all add up, the total error tends to form a bell shape. This is the kind of situation the lesson describes when it says that sums of many independent pieces look normal.

3 checked practice questions for this part are in the app.

Part 5 · 40:00–50:28

This part explains why normal random variables stay normal under linear changes, and how to use a single standard normal table to find probabilities for any normal.

Key ideas

If X is normal, then any linear function of X (multiplying by a constant and adding a constant) is also normal. The lecturer says the proof comes in a few lectures. 40:00
The normal curve's area has no simple formula, so people calculate its values once and put them in a table. Only one table is needed, the one for the standard normal, with mean 0 and variance 1. 42:06
To use the table for any normal, you standardize it: subtract the mean, then divide by the standard deviation. The result is a standard normal. 44:41
In the example, a normal with mean 2 and variance 16 gives P(X less than 3) equal to P(standard normal less than 0.25), which the table gives as 0.987. 46:55
Continuous random variables use densities, which work like the probability mass function in the discrete case. A density is a rate at which probability builds up, not a probability itself, and continuous cases use integrals where discrete cases use sums. 49:03

Words to know

standard normal: A normal bell curve with average 0 and variance 1. A table lists its values, so you can use it for any normal.
standardizing: Changing a random variable by subtracting its mean and dividing by its standard deviation, so it has mean 0 and variance 1.
closed form: A formula written with ordinary math operations that gives the answer directly. The normal integral does not have one.
cumulative distribution function (CDF): A function that gives the probability that a random variable is below a certain value.

Example

Think of a quiz score that is 3 standard deviations above the class average. The lecturer says this tells you how many standard deviations you are from the mean, which is a fair way to compare results from different tests. Standardizing turns any normal result into this same kind of score, so one table can answer every question.

3 checked practice questions for this part are in the app.

← Lecture 5: Discrete Random Variables; Probability Mass Functions; Expectations · Lecture 12: Iterated Expectations; Sum of a Random Number of Random Variables →

Notes written by AI (Claude) from the lecture's official transcript, so check anything important against the lecture. MIT OpenCourseWare · CC BY-NC-SA 4.0 · not affiliated with or endorsed by MIT. These notes are shared under the same license.

LessonBound · Free lecture notes · Privacy · Terms · New York