Introduction and Prerequisites

• Educational video by Josh Starmer on StatQuest
• Topic: Expected values for continuous variables (Part 2)
• Assumes familiarity with expected values Part 1 (discrete variables)
• References previous example: Statsquatch's bet about Troll 2 movie
  • Expected value was 0.66 (gain 66 cents per bet on average)

Discrete vs Continuous Variables Definition

• Discrete variables: Have discrete outcomes (like betting - lose $1 or gain $1)
• Continuous variables: Come from measuring things, outcomes are continuous
• Example scenario: Measuring waiting time between meeting people in Statland

Data Collection and Modeling

• Initial data collection: Recording wait times (10, 30, 0, 10 seconds, etc.)
• Problem with discrete intervals: Gaps in data, interval size concerns
• Solution: Model with exponential distribution
• Exponential distribution equation: f(x) = λe^(-λx)
  • λ (lambda) = rate parameter (people met per second)
  • Defines curve shape

Exponential Distribution Parameters

• λ = 0.05: Fits collected data
• λ = 0.1: Steeper slope, more people per second (green curve)
• λ = 0.01: Flatter curve, fewer people per second (orange curve)
• Curve extends theoretically to positive infinity

Probability Calculations

• Probability of meeting someone ≤10 seconds: Integrate from 0 to 10
  • Result: 0.39 when λ = 0.05
• Probability of meeting someone between 25.302-30.122 seconds: 0.06
• Y-axis represents likelihood values (used in maximum likelihood estimation)
• Total area under curve = 1

Approximating Continuous with Discrete

• Divide time into 10-second intervals (rectangles)
• Rectangle height = exponential function value at midpoint
• Rectangle area = height × width = approximate probability
• First interval (0-10 seconds):
  • Height at x=5: 0.04
  • Area: 0.04 × 10 = 0.4
  • Compare to exact integral: 0.39

Expected Value Calculation - Discrete Approximation

• Using 10-second intervals:
  • First term: 10 × 0.4
  • Second term: 20 × 0.2
  • Result: 22 seconds average wait time
• Using 5-second intervals: 21.8 seconds (better approximation)
• As interval width → 0 and rectangles → infinity: exact calculation

Continuous Expected Value Formula

• Discrete: E[X] = Σ(outcomes × probabilities)
• Continuous: E[X] = ∫x·f(x)dx
• Key differences:
  1. Replace sum with integral
  2. Replace probability with likelihood (y-axis coordinate)
• Formula works for any continuous distribution

Mathematical Derivation for Exponential Distribution

• E[X] = ∫₀^∞ x·λe^(-λx)dx
• Integration by parts:
  • f(x) = x, f'(x) = 1
  • g'(x) = λe^(-λx), g(x) = -1/λ·e^(-λx)
• L'Hôpital's rule applied for limit as x→∞
• Final result: E[X] = 1/λ

Final Calculation and Answer

• Given λ = 0.05
• Expected waiting time = 1/0.05 = 20 seconds
• Answer to Statsquatch's question: Average 20 seconds between meeting people

Key Takeaways

• Expected values for discrete vs continuous variables very similar
• Two main differences:
  1. Sum → Integral
  2. Probability → Likelihood
• Formula: E[X] = 1/λ for exponential distribution