Discrete Random Variables, Expected Value, and Probability Distributions

Fundamentals of Random Variables and Variable Types

  • Random Variable Definition and Notation:

    • A random variable is a variable whose possible values are numerical outcomes of a random phenomenon.
    • Random variables are denoted using capital letters, such as XX.
    • Specific numerical values that a random variable can assume are denoted using lowercase letters, such as xx.

  • Discrete Random Variables:

    • A discrete random variable has either a finite or a countable number of values.
    • Countable values imply data obtained through counting (e.g., 0,1,2,3,40, 1, 2, 3, 4).
    • On a standard number line, the values of a discrete random variable are plotted as individual, isolated points with gaps in between them.
    • Example 1: A discrete random variable taking on whole number values between 00 and 44 (x×values×are×0,1,2,3,4x \times \text{values} \times \text{are} \times 0, 1, 2, 3, 4). On a number line, these are plotted strictly as distinct points at 00, 11, 22, 33, and 44.
    • Example 2: Count data where values begin at 0, 1, 2, 3, 4, 5, 7, 8, 9, 9, 9\n\n- **Continuous Random Variables**:\n - A continuous random variable can take on any real value within a continuous interval on a number line.\n - Continuous data is measured rather than counted.\n - On a number line, a continuous random variable is plotted as an uninterrupted line across an interval because it assumes any real number in that range.\n - Example 1: Any decimal or fractional value between two numbers, such as values between 1.4andand2.7,oranyrealnumberbetween, or any real number between0andand4.\n - Example 2: The length of time between calls to 911. The lower bound for time is fixed at 0(timemustbegreaterthan(time must be greater than0),andwhiletheupperboundisunknown,timecantakeonanycontinuousdecimalvaluegreaterthan), and while the upper bound is unknown, time can take on any continuous decimal value greater than0.\n\n# Characteristics and Rules of Discrete Probability Distributions\n\n- **Definition of Probability Distribution**:\n - A probability distribution provides all possible values that a random variable can assume along with the corresponding probability for each of those values.\n - It represents every outcome that can possibly happen in a given random process and the exact likelihood of each option.\n\n- **Probability Distribution vs. Relative Frequency Distribution**:\n - *Relative Frequency Distribution*: Based on empirical sample data collected from a specific sample data set. It only includes outcomes present in that specific sample data set and does not necessarily capture every outcome that could theoretically happen.\n - *Probability Distribution*: Encompasses **every possible outcome** in the entire sample space along with its theoretical or population probability.\n\n- **Tabular Presentation**:\n - Discrete probability distributions are commonly presented in tables containing values of the random variable xandtheircorrespondingprobabilitiesand their corresponding probabilitiesP(X = x).\n\n- **Rules for Discrete Probability Distributions**:\n - To verify whether a given table or chart is a valid discrete probability distribution, ignore the x values and evaluate the probabilities against two key rules:\n 1. **Sum of Probabilities Rule**:\n     \sum P(x) = 1\n     The sum of all probabilities in a discrete probability distribution must equal exactly 1.Becauseaprobabilitydistributionaccountsforeverypossibleoutcomethatcanoccur,itstotalprobabilityisequivalenttotheprobabilityoftheentiresamplespace,whichalwaysequals. Because a probability distribution accounts for every possible outcome that can occur, its total probability is equivalent to the probability of the entire sample space, which always equals1\n 2. **Probability Bounds Rule**:\n     0 \le P(x) \le 1\n     Each individual probability P(x)mustbearealnumbergreaterthanorequaltomust be a real number greater than or equal to0andlessthanorequaltoand less than or equal to1.\n\n# Measures of Central Tendency and Law of Large Numbers\n\n- **Population Mean of a Discrete Probability Distribution (\mu)**:\n - The mean of a discrete probability distribution represents the population mean (\mu).\n - It is calculated as a weighted mean where the values of the random variable (x)serveasthevaluesandthecorrespondingprobabilities() serve as the values and the corresponding probabilities (P(x)) serve as the weights.\n - *General Weighted Mean Formula*:\n    \mu = \frac{\sum (x \times w)}{\sum w}\n - *Discrete Probability Distribution Mean Formula*:\n    \mu = \sum (x \times P(x))\n - *Omission of Division by Sum of Weights*: In the general weighted mean formula, the expression is divided by the sum of the weights (\sum w).Inadiscreteprobabilitydistribution,theweightsareprobabilities,andbydefinition,thesumofallprobabilitiesequals). In a discrete probability distribution, the weights are probabilities, and by definition, the sum of all probabilities equals1((\sum P(x) = 1).Dividingby). Dividing by1doesnotchangethenumerator,simplifyingtheformuladirectlytodoes not change the numerator, simplifying the formula directly to\sum (x \times P(x)).\n\n- **Law of Large Numbers**:\n - As the sample size (n)increases() increases (n \rightarrow \infty),thesamplemean(), the sample mean (\bar{x})calculatedfrom) calculated fromntrialsapproachesthetheoreticalpopulationmean(trials approaches the theoretical population mean (\mu).\n - The absolute difference between the sample mean and population mean approaches zero as n grows larger:\n    |\bar{x} - \mu| \rightarrow 0 \quad \text{as} \quad n \rightarrow \infty\n - A small sample size may produce a sample mean that deviates significantly from the population mean. However, as the number of trials increases, the sample mean becomes increasingly accurate.\n\n# Concept and Calculation of Expected Value\n\n- **Expected Value Terminology and Notation**:\n - Denoted as E(X)ororE(x).\n - Represents the theoretical long-run average outcome of a random variable over repeated trials.\n - Originates historically from games of chance and gambling, where players calculated the expected financial outcome in terms of money over the long run.\n - Widely used in financial, business, insurance, and risk management evaluation.\n\n- **Equivalence to Mean**:\n - The expected value E(X)ismathematicallyidenticaltothepopulationmean(is mathematically identical to the population mean (\mu):\n    E(X) = \mu = \sum (x \times P(x))\n - While the terminology changes to *expected value* in financial contexts, the underlying mathematical calculation remains identical to finding a population mean.\n\n- **Perspective / Point of View Requirement**:\n - Expected value problems in financial settings must be calculated from a clearly specified perspective (point of view).\n - The expected financial outcome for a buyer or client is the exact opposite of the expected financial outcome for a seller, casino, or insurance company.\n\n- **Net Gain as the Random Variable (X)**:\n - In financial expected value problems, the random variable X represents **Net Gain**.\n - Net Gain is defined as the net difference between initial monetary position and final monetary position:\n    \text{Net Gain} = \text{Ending Amount} - \text{Starting Amount}\n - Net Gain values can be positive (profit), negative (loss), or zero (break-even).\n\n# Comprehensive Case Study: Life Insurance Expected Value Analysis\n\n- **Scenario Parameters**:\n - Subject: A 49-year-old female.\n - Insurance policy duration: 1 year.\n - Policy purchase price (Premium): \$530.\n - Death benefit payout (Coverage): \$250{,}000.\n - Mortality probability (P(\text{Die})):):0.00209.\n - Perspective: The Insurance Company.\n\n- **Constructing the Probability Distribution**:\n - Over the course of the 1-year policy term, exactly two outcomes are possible for the policyholder: she lives or she dies.\n - *Outcome 1: Policyholder Lives*:\n - The insurance company starts at \$0,receives, receives\$530forthepolicy,andpaysoutfor the policy, and pays out\$0 at the end of the year.\n - Net Gain for insurance company (x_1):):+\$530\n - Probability (P(x_1)):Calculatedusingtherulethattotalprobabilitiesmustsumto): Calculated using the rule that total probabilities must sum to1:\n      P(\text{Live}) = 1 - P(\text{Die}) = 1 - 0.00209 = 0.99791\n - *Outcome 2: Policyholder Dies*:\n - The insurance company starts at \$0,receives, receives\$530forthepolicy,andmustpayfor the policy, and must pay\$250{,}000 to the beneficiary upon death.\n - Net Gain for insurance company (x_2):):\$530 - \$250{,}000 = -\$249{,}470\n - Probability (P(x_2)):):0.00209\n - *Probability Distribution Table Summary*:\n - x_1 = +\$530withwithP(x_1) = 0.99791\n - x_2 = -\$249{,}470withwithP(x_2) = 0.00209\n - Probability check: 0.99791 + 0.00209 = 1.00000\n\n- **Calculation of Insurance Company Expected Value**:\n - Apply the expected value formula E(X) = \sum (x \times P(x)):\n    E(X) = (530 \times 0.99791) + (-249470 \times 0.00209)\n    E(X) = 528.8923 + (-521.3923) = +\$7.50\n\n# Long-Run Interpretation and Practical Applications of Expected Value\n\n- **Interpretation of Sign in Expected Value**:\n - **Positive Expected Value (E(X) > 0)**: Indicates long-term profitability or financial gain.\n - **Negative Expected Value (E(X) < 0)**: Indicates long-term financial loss.\n - **Zero Expected Value (E(X) = 0)**: Indicates a fair situation where no party gains or loses on average.\n\n- **Long-Run Average vs. Individual Trial Outcomes**:\n - An expected value of \$7.50 is a long-term **average** per policy across many policies; it is not the actual outcome of any single individual policy.\n - For a single policy sold (n = 1),thecompanynevermakes), the company never makes\$7.50;iteithermakes; it either makes\$530(ifthewomanlives)orloses(if the woman lives) or loses\$249{,}470 (if the woman dies).\n - If the company sells 1{,}000{,}000policies(policies (n = 1{,}000{,}000):\n - Most women live, allowing the company to retain \$530 per policy.\n - A small fraction die, requiring \$250{,}000 payouts.\n - On average, the company makes \$7.50 per policy sold.\n - Total expected net gain across all policies: 1{,}000{,}000 \times \$7.50 = \$7{,}500{,}000.\n\n- **Casino Games and Gambling Applications**:\n - From the player's perspective, all casino games have a negative expected value (E(X) < 0).\n - Playing casino games over a long period guarantees net financial loss for the player.\n - Multi-outcome events (such as roulette, where outcomes include winning on a specific number, winning on a color, or losing) are calculated by identifying all potential outcomes, finding net gain for each, ensuring all probabilities sum to 1,andcomputing, and computing\sum (x \times P(x)).\n\n# Variance and Standard Deviation of Discrete Random Variables\n\n- **Notation**:\n - Population Variance is denoted by \sigma^2\n - Population Standard Deviation is denoted by \sigma\n\n- **Formulas for Standard Deviation**:\n - Definitional Formula:\n    \sigma = \sqrt{\sum ((x - \mu)^2 \times P(x))}\n - Computational Formula (Algebraic equivalent, easier to compute):\n    \sigma = \sqrt{\sum (x^2 \times P(x)) - \mu^2}\n\n- **Formula for Variance**:\n - Squaring both sides eliminates the radical, giving the population variance formula:\n    \sigma^2 = \sum (x^2 \times P(x)) - \mu^2\n\n- **Critical Computational Rule and Common Error**:\n - The most frequent calculation error is forgetting to subtract the squared population mean (\mu^2)fromthesumofproducts() from the sum of products (\sum (x^2 \times P(x))).\n - Always calculate the variance underneath the square root symbol first before taking the square root.\n\n# Step-by-Step Computational Example for Variance and Standard Deviation\n\n- **Context and Parameters**:\n - Data context: Number of movies streamed in a day discrete probability distribution.\n - Calculated Population Mean: \mu = 1.49\n - Squared Mean: \mu^2 = (1.49)^2\n\n- **Calculator Optimization Strategy**:\n - Simplify values before typing full expressions into a calculator to prevent exceeding character limits or memory constraints:\n - Compute squares of xfirst:first:0^2 = 0,,1^2 = 1,,2^2 = 4,,3^2 = 9, etc.\n - Eliminate zero terms: 0^2 \times P(0) = 0\n - Simplify unit terms: 1^2 \times 0.58 = 1 \times 0.58 = 0.58\n\n- **Step 1: Compute Variance Underneath the Radical**:\n  \sigma^2 = \sum (x^2 \times P(x)) - \mu^2\n  \sigma^2 = (0^2 \times P(0)) + (1^2 \times 0.58) + (2^2 \times P(2)) + (3^2 \times P(3)) + \dots - (1.49)^2\n  \sigma^2 = 0.869\n\n- **Step 2: Compute Standard Deviation**:\n - Take the square root of the computed variance:\n    \sigma = \sqrt{\sigma^2} = \sqrt{0.869}$$