Discrete vs Continuous Distributions

A **probability distribution** describes how likely different outcomes are for a random variable. ## Discrete Distributions A **discrete distribution** applies to variables that can only take **specific, separate values** (usually counted). **Examples of discrete variables:** - Number of heads in 10 coin flips: $0, 1, 2, \ldots, 10$ - Roll of a die: $1, 2, 3, 4, 5, 6$ - Number of students in a class: $15, 16, 17, \ldots$ **Key feature:** You can list all possible values. We use a **Probability Mass Function (PMF)** where: $$P(X = x) \geq 0 \text{ and } \sum_{\text{all } x} P(X = x) = 1$$ ## Continuous Distributions A **continuous distribution** applies to variables that can take **any value in a range** (usually measured). **Examples of continuous variables:** - Height of a person: any value from 0 to 300 cm - Time to complete a task: any positive value - Temperature: any value on a thermometer **Key feature:** You cannot list all possible values (infinitely many). We use a **Probability Density Function (PDF)** where: $$P(a \leq X \leq b) = \int_a^b f(x)\, dx$$ For continuous variables, $P(X = \text{exact value}) = 0$ because there are infinitely many possible values.

Classifying Random Variables

Classify each variable as discrete or continuous: (a) Number of cars passing a checkpoint (b) Weight of a package (c) Number of defective items in a batch

Analyze variable (a): Cars are counted: 0, 1, 2, 3, ... Cannot have 2.5 cars = Discrete

Analyze variable (b): Weight is measured: 1.234 kg, 1.235 kg, etc. Any value possible = Continuous

Analyze variable (c): Defects are counted: 0, 1, 2, ... Cannot have 1.7 defects = Discrete

Answer: (a) Discrete, (b) Continuous, (c) Discrete

Calculating Discrete Probabilities

A fair die is rolled. Find: (a) $P(X = 4)$ and (b) $P(X \leq 2)$

Identify the distribution: Die roll has 6 equally likely outcomes: 1, 2, 3, 4, 5, 6 = Uniform discrete distribution

Calculate P(X = 4): Only one outcome equals 4, so $P(X = 4) = \frac{1}{6}$ = $P(X = 4) = \frac{1}{6}$

Calculate P(X ≤ 2): Two outcomes satisfy this: 1 and 2 $P(X \leq 2) = P(X=1) + P(X=2) = \frac{1}{6} + \frac{1}{6}$ = $P(X \leq 2) = \frac{2}{6} = \frac{1}{3}$

Answer: (a) $\frac{1}{6} \approx 0.167$, (b) $\frac{1}{3} \approx 0.333$

Understanding Continuous Probability

Heights in a population follow a continuous distribution. Explain why $P(\text{height} = 170.0000\ldots \text{ cm})$ equals zero.

Consider the nature of continuous variables: Height can be 170.0001 cm, 170.00001 cm, etc. Infinitely many values exist = Infinite possible values

Apply probability reasoning: If each exact value had positive probability, the sum would be infinite (not 1) = Each exact value must have probability 0

Understand the practical meaning: We calculate $P(169.5 < X < 170.5)$ instead. This is the area under the curve between these limits = Use intervals, not exact values

Answer: With infinitely many possible values, assigning positive probability to each would make the total exceed 1. For continuous distributions, we calculate probabilities for intervals, not exact values.

Understanding the difference between discrete and continuous distributions is fundamental to: - **Statistics**: Choosing the right test or model for your data - **Data Science**: Selecting appropriate machine learning algorithms - **Quality Control**: Monitoring manufacturing processes (defect counts vs measurements) - **Finance**: Modeling stock prices (continuous) vs number of trades (discrete) - **Medicine**: Analyzing patient counts (discrete) vs blood pressure readings (continuous) Using the wrong type of distribution leads to incorrect conclusions and predictions!

Mistake: Thinking 'money' is always discrete because we count dollars

Why: While we often round money to cents, financial models frequently treat money as continuous because transactions can theoretically be any amount.

Correct: Context matters: counting coins is discrete, but modeling stock prices is typically continuous.

Mistake: Trying to calculate P(X = exact value) for continuous distributions

Why: For continuous distributions, the probability at any single point is zero. Only intervals have positive probability.

Correct: For continuous distributions, always calculate $P(a < X < b)$ using the area under the curve.

Mistake: Confusing 'few possible values' with 'discrete'

Why: A variable is discrete based on its nature, not the number of values observed in a sample.

Correct: Ask: 'Can this variable theoretically take ANY value in a range?' If yes, it's continuous.

Quality Control in Manufacturing

Factories use discrete distributions to count defects and continuous distributions for measurements.

A factory counts defective items per batch (discrete: 0, 1, 2, ...) but measures product weight in grams (continuous: 250.3g, 250.31g, ...).

Medical Statistics

Healthcare uses both types: patient counts (discrete) and vital signs (continuous).

Number of patients admitted today (discrete: 45, 46, 47) vs blood pressure reading (continuous: 120.5 mmHg, 121.3 mmHg).

**Discrete distributions** apply to variables with countable, separate values (like dice rolls, coin flips, or counts)

**Continuous distributions** apply to variables that can take any value in a range (like height, weight, or time)

For discrete variables, we use PMF and can calculate $P(X = x)$ directly

For continuous variables, we use PDF and calculate probabilities for intervals: $P(a < X < b)$

Ask yourself: 'Is this counted or measured?' Counted = discrete, Measured = continuous

Q: Can a variable be both discrete and continuous?

A: Usually a variable is one or the other. Some variables are mixed: daily rainfall is exactly 0 mm on dry days (a positive probability for one exact value) but can be any amount on rainy days. We also sometimes model discrete variables (like large counts) with continuous distributions as an approximation.

Q: Why is age sometimes treated as discrete and sometimes continuous?

A: Age in years (18, 19, 20) is discrete. Age as exact time since birth (18.5273 years) is continuous. The choice depends on how you measure and use the data.

Q: What is the difference between PMF and PDF?

A: PMF (Probability Mass Function) gives the actual probability for discrete values. PDF (Probability Density Function) gives density for continuous variables - you must integrate over an interval to get probability.

Exit Presentation

Discrete vs Continuous Distributions

1 / 11

Probability / Probability Distributions

Discrete vs Continuous Distributions

Learn the difference between discrete and continuous probability distributions, and when to use each type.

Advanced25 min

Use arrow keys or spacebar to navigate