Comprehensive Notes on Continuous Probability Distributions and the Normal Distribution

Comparison of Discrete and Continuous Probability Distributions

• Conceptual Distinction

  • Discrete Variables (Chapter 5 Recap): Involve countable outcomes that are integer values. A key example is the number of people in a room; it is impossible to have values like $1.5$ or $2.5$ people. On a number line, outcomes (e.g., $1$, $2$, and $3$) are distinct points.

  • Continuous Variables (Chapter 6): Involve uncountable outcomes that can take on any real value within a range. For instance, on a number line between $1$ and $3$, a variable could be exactly $1.5$, $1.6$, or even $1.999999999$. The possibilities are effectively endless because the range can be partitioned into infinitely small segments. These typically involve measurements such as volume, height, length, and time.

• Function Types

  • Probability Mass Function (PMF): Used for discrete distributions. On a graph where the x-axis represents outcomes ($x$) and the y-axis represents probability, the probability is equal to the height of the line at a specific point.

  • Probability Density Function (PDF): Used for continuous distributions. On a graph where the x-axis represents outcomes ($x$) and the y-axis represents a function of x ($f(x)$), the graph takes on a specific shape (e.g., a bell curve). Probability is no longer height; it is measured as the area under the curve between two points ($a$ and $b$).

• Mathematical Rules

  • Discrete Rule: The sum of all probabilities in the PMF must equal one: $\sum P(x) = 1$.

  • Continuous Rule: The entire area under the PDF curve must equal one. Mathematically, the integral of the function from negative infinity to infinity equals one: $\int_{-\infty}^{\infty} f(x) \, dx = 1$.

• Specific Point Probabilities

  • In discrete distributions, we can calculate probabilities at exact points, such as $P(X = 5)$.

  • In continuous distributions, the probability at an exact point is always zero: $P(X = 5) = 0$. Because the range is infinite and we require area (width times height) to find probability, a single vertical line has no width and thus no area. Instead, we calculate probabilities over intervals, such as P(4.5 < X < 5.5), to look at the "neighborhood" of a value.

The Normal Distribution

• Significance and Occurrence

  • The normal distribution is considered a "rule of the universe" because it emerges frequently in real-world data across various fields. If plotted on a histogram, many datasets naturally form a symmetric bell shape.

  • It is the foundation of the Central Limit Theorem, which suggests that the distribution of sample means will tend toward a normal distribution regardless of the original population's distribution shape.

• Theoretical Properties and Notation

  • Shape: Symmetric bell-shaped curve with high density in the center and fading tails.

  • Notation: $X \sim N(\mu, \sigma^2)$.

  • Parameters:

    • The Mean ($\mu$): Determines the center of the distribution. Shifting the mean moves the entire curve along the x-axis (e.g., from $10$ to $20$ to $30$).

    • The Variance ($\sigma^2$) or Standard Deviation ($\sigma$): Determines the variability or width of the curve. A larger standard deviation results in a wider, flatter curve, while a smaller standard deviation creates a narrower, taller curve. Standard deviation measures the tendency of data to deviate from the mean.

    • Conversion: To get standard deviation from variance, find $\sqrt{\sigma^2}$. To get variance from standard deviation, find $\sigma^2$.

• The Probability Density Function (PDF) Formula

  • The mathematical form governing the bell shape is: $f(x) = \frac{1}{\sigma \sqrt{2\pi}} e^{-\frac{1}{2}\left(\frac{x - \mu}{\sigma}\right)^2}$.

  • Components: Includes the mathematical constant $\pi$ ($3.14159$) and the natural base $e$. Students are required to recognize this formula as the normal distribution PDF but generally do not solve it by hand.

Standardizing the Distribution (Z-Scores)

• Integration and Closed Form Solutions

  • Finding the area under the normal curve theoretically requires integration: $\int_{a}^{b} f(x) \, dx$. However, this integral has no closed-form solution (it cannot be solved with standard calculus formulas). It must be solved numerically via computers or calculators.

• The Standard Normal Distribution ($Z$)

  • To simplify calculations, every normal distribution can be converted to the "Universal Scale" known as the Standard Normal Distribution.

  • Notation: $Z \sim N(0, 1)$, where the mean is always $0$ and the standard deviation is always $1$.

  • Transformation Formula: $z = \frac{x - \mu}{\sigma}$.

Calculating Probabilities Using the HP Calculator

• The Procedure

  • Step 1: Convert the given $x$ value to a $z$ value using the standardization formula.

  • Step 2: Use the calculator to find the probability $P(X \le x)$. The HP calculator is designed specifically to work with "less than" ($\le$) probabilities.

• Calculator Key Sequences

  • To find probability from a Z-value:

    1. Enter the $z$ value (e.g., $0.12$).

    2. Press the Blue Shift button.

    3. Press button 3 (indexed as "zed to p" or $Z \rightarrow P$).

  • To adjust decimal display:

    1. Press the Orange Shift button (downward arrow).

    2. Press the Equals ($=$, Display) button.

    3. Type the desired number of decimals (e.g., $4$).

• Handling "Greater Than" (>) Problems

  • A "greater than" sign is a "reason for pause" because the calculator only provides "less than" area.

  • Complement Rule: Since the total area is $1$, the probability of the right tail is calculated as: P(X > x) = 1 - P(X < x).

• Equality Notation in Chapter 6

  • In discrete distributions (Chapter 5), the difference between < and $\le$ is critical.

  • In continuous distributions (Chapter 6), they are effectively the same because the probability at any specific point is zero. Thus, P(X < 18) is indistinguishable from $P(X \le 18)$.

Example Case Study: Image Download Time

• Scenario: $X$ represents the time in seconds to download an image file. $X$ is normally distributed with $\mu = 18$ seconds and $\sigma = 5$ seconds.

• Problem A: Calculate P(X < 18.6)

  1. Convert to Z: $z = \frac{18.6 - 18}{5} = \frac{0.6}{5} = 0.12$.

  2. Calculator Entry: Input $0.12$, press Blue Shift, then 3.

  3. Result: P(X < 18.6) = 0.5478. This represents the area to the left of $18.6$ on the distribution curve.

• Problem B: Calculate P(X > 18.6)

  1. Apply Complement Rule: P(X > 18.6) = 1 - P(X < 18.6).

  2. Substitution: $1 - 0.5478 = 0.4522$.

Questions & Discussion

• Question: How did you get more decimals on the calculator?

• Response: Use the Orange Shift button, then the Equals button (which corresponds to "Display" in orange text), and then hit the number corresponding to the decimals you want (usually $4$).

• Question/Note on Tables: The lecturer demonstrated a standard $Z$ table, showing that for $z = 0.12$, you find the intersection of row $0.1$ and column $0.02$ to get $0.5478$.

• Note Regarding Examinations: Traditional printed $Z$ tables are being phased out. Students are explicitly required to have and use the HP calculator for exams, as it is also necessary for the second-semester course, Theory of Interest.