Module 2: Probability and Probability Distributions — Notes
Lesson 1: Learning Objectives
Module Overview:
The module focuses on understanding probability theory and probability distributions with the following key learning objectives:
Learning Objectives
- Explain probability theory, types and algebra of events, rules of probability
- Discuss conditional, joint and marginal probability and Bayes' theorem
- Describe probability distribution, variable types and distribution basics and types
- Learn about Poisson distribution, distribution and types of distribution
- Understand relevant industry example and case study
Lesson 2: 2.1_Probability Theory
Sub-lesson 2.1.1: Introduction of Probability
Key Definitions:
Probability is a quantitative measure of risk.
Per Statistician I. J. Good: "The theory of probability is much older than the human species, since the assessment of uncertainty incorporates the idea of learning from experience, which most creatures do."
Reading Content (SLM Pages 44-45):
Sampling and probability are integral components of statistics. We need to be familiar with a few common terminologies used in probability theory before we can discuss sampling distributions and probability. Despite being frequently used in business contexts, these phrases have precise technical meanings.
Key Terminology and Definitions (Accordion Sections):
-
Random Experiment: A procedure or action that yields findings under investigation is referred to as an experiment in probability theory. An example would be sampling from a production lot. An experiment with a random outcome is one whose results are unknown beforehand. Every outcome carries a chance or risk, which is also frequently referred to as uncertainty.
-
Sample Space: An experiment's set of all potential results. Typically, S is used to symbolise it. - Example: If the random experiment is rolling of a die, the sample space is: S = {1, 2, 3, 4, 5, 6} - Example: If the random experiment is tossing of three coins, the sample space is: S = {HHH, HHT, HTH, THH, HTT, THT, TTH, TTT} with total of 8 possible outcomes (H = heads, T = tails) - Example: Quality inspection - if we randomly select two products from a production lot and inspect them for defects: S = {DD, DS, DR, RS, RR, SS}, where D = faulty, S = serviceable, R = re-workable
-
Event: An event is one or more potential outcomes that fit into a particular interest category. An event is a subset E of the sample space S. Put otherwise, an event is a successful result.
-
Event Space: All potential events are contained within it. Typically, it is shown as E. Keep in mind that the number of elements in the sample space and the number of elements in the event space are often of importance in probability and statistics.
-
Union of Events: Given two events, E and F, an additional event is defined as the set of all possible outcomes that fall into either E, F, or both. This additional event is referred to as the union of events E and F. It has the notation E U F.
-
Intersection of Events: When two events, E and F, are defined as such, an additional event that encompasses all possible outcomes from both E and F is referred to as the intersection of events E and F. It is written as E ∩ F.
-
Mutually Exclusive Events: If there isn't a common experiment conclusion between events E and F, then they are considered mutually exclusive occurrences. Stated differently, if E ∩ F = φ, where φ is a null or empty set, then events E and F are mutually exclusive.
-
Collectively Exhaustive Events: If the sample space equals the union of the events, then the events are collectively exhaustive.
-
Event Complement: All of the outcomes that are not included in the E make up the event E. It has the symbol EC. E U EC = S and E ∩ EC = φ follow.
Quiz Questions: - Q1: "Probability is defined as:" → Correct Answer: "The likelihood of an event occurring" - Q2: "The range of probability values is:" → Correct Answer: "0 to 1"
Sub-lesson 2.1.2: Types of Events
Reading Content (SLM Pages 45-46):
A probability event can be defined as a set of outcomes from an experiment. In other words, an event in probability is the subset of the respective sample space. A random experiment's entire potential set of outcomes is the sample space or the individual space of that encounter. The probability of an occurrence happening is called chance. The likelihood of any event happening lies between 0 and 1.
Example: The sample space for the tossing of three coins simultaneously is given by: S = {(T, T, T), (T, T, H), (T, H, T), (T, H, H), (H, T, T), (H, T, H), (H, H, T), (H, H, H)}
Suppose if we want to find only the outcomes which have at least two heads; then the set of all such possibilities can be given as: E = {(H, T, H), (H, H, T), (H, H, H), (T, H, H)}
Thus, an event is a subset of the sample space, i.e., E is a subset of S.
There could be a lot of events associated with a given sample space. For any event to be recognized as an "event", it needs to have at least one or more outcomes from the sample space.
Events may include: - Independent (none of the occurrences affects the others). - Dependent (also known as 'Conditional'): an event that is influenced by other events. - Mutually exclusive (no two things may occur simultaneously). - An event occurs when an "even number"—2, 4, or 6—is rolled.
Quiz Questions: - Q1: "Mutually exclusive events are:" → Correct Answer: "Events where one occurrence prevents the other" - Q2: "Independent events imply:" → Correct Answer: "No influence on each other's occurrence"
STATUS NOTE: Lesson 2.1.2 completed successfully with both quizzes answered correctly. The platform lock encountered at the time was later confirmed (2026-09-25) to be a transient one-off video-player glitch, resolved on a fresh-session retry — see PROGRESS.md for detail. Lesson 4 is fully complete.
Lesson 5: 2.1.3_Algebra of Events
Reading Content (SLM Pages 46-47):
Events are the outcome of an experiment. The likelihood of an event occurring is the ratio of the number of favourable events to the total number of occurrences. Often, two things occur together, or it can happen that just one of them occurs.
Event algebra can offer an event that performs certain operations over two given events. The operations are union, intersection, complement and difference of two events. As events are subsets of sample space, these operations are performed as set operations.
Complementary Events
For an event A, there is a complementary event B such that B represents the set of outcomes which are not in the set A.
Example: If two coins are tossed together, the sample space is S = {HT, TH, HH, TT}. Let A be the event of getting one head: A = {HT, TH}. The complementary event, B = {HH, TT}.
Events with AND (Intersection)
AND stands for the intersection of two sets. An event is the intersection of two events if it has members present in both events.
Example: If a pair of dice is rolled, the sample space has 36 members. Suppose A is the event of getting both dice showing the same number, and B is the event having the sum as 6. - A = {(1,1), (2,2), (3,3), (4,4), (5,5), (6,6)} - B = {(3,3), (1,5), (5,1), (2,4), (4,2)} - A AND B = {(3,3)}
Events with OR (Union)
OR stands for the union of two sets. An event is called the union of two events if it has members present in either of the sets.
Example: If two coins are tossed together, S = {HT, TH, TT, HH}. Let event A be the event having only one head and event B be the event having two heads. - A = {HT} - B = {HH} - Union of A and B, A OR B = {HT, HH}
Events with BUT NOT (Difference)
For two events A and B, "A but not B" is the event having all the elements of A but excluding the elements of B. This can also be represented as A − B.
Example: Suppose there is an experiment of choosing 4 cards from a deck of 52 cards. Event A is having all cards as red cards, and event B is having all cards as kings. Then event "A but not B" will have all red cards excluding the two red kings.
Quiz Questions: - Q1: "Union of events A and B (A ∪ B) means:" → Correct Answer: "A or B or both occur" - Q2: "Intersection of events A and B (A ∩ B):" → Correct Answer: "Both A and B occur simultaneously"
Set-operation vocabulary for events: AND = intersection (∩), OR = union (∪), BUT NOT = difference (−), complement = "not A". Mapping these English connectives to their set-notation equivalents is the recurring theme tested in this lesson's quiz.
Technical note: this lesson's video is an animated-infographic-style Vimeo embed whose static first frame visually resembles a content slide (no obvious "play" triangle overlay). The actual play button only becomes visible after clicking once into the video area (reveals the control bar with a small play icon, bottom-left) — don't mistake the static poster-frame infographic for a non-video content block. See PROGRESS.md Technical Findings for the full diagnostic detail.
Lesson 6: 2.1.4_Addition Rule of Probability
Reading Content (SLM Pages 47-48):
There are (n₁ + n₂) ways to complete one of two jobs (either one task or the other) if one task can be completed in n₁ ways and another in n₂ ways, when the two tasks cannot be completed simultaneously. The "Sum Rule" applies when using logical OR to determine an experiment's outcomes, and when occurrences are mutually exclusive.
The Addition Rule of Probability states that:
- If A and B are any two events, the probability of the occurrence of either A or B is given by: P(A ∪ B) = P(A) + P(B) − P(A ∩ B) (General Addition Rule)
- If A and B are two mutually exclusive events, the probability of occurrence of either A or B is given by: P(A ∪ B) = P(A) + P(B) (applicable only when A ∩ B = 0)
Extension to 3 events: P(A ∪ B ∪ C) = P(A) + P(B) + P(C) − P(A∩B) − P(A∩C) − P(B∩C) + P(A∩B∩C) — used in complex situations involving three overlapping events.
Worked Example 1: An urn contains 10 balls: 5 white, 3 black, 2 red. If we select one ball randomly, how many ways are there that the ball is either white or red? Answer: 5 + 2 = 7 (mutually exclusive — simple sum rule).
Worked Example 2 (solved via accordion "Solution"): In a triangular series, there is a 0.7 chance India wins against Zimbabwe and a 0.4 chance against Australia, with a 0.3 chance of winning both. What is the likelihood India wins at least one match? - P(A) = 0.7 (vs Zimbabwe), P(B) = 0.4 (vs Australia), P(A ∩ B) = 0.3 - P(A ∪ B) = P(A) + P(B) − P(A ∩ B) = 0.7 + 0.4 − 0.3 = 0.8
Real-world applications covered: Venn diagrams as visual tools; insurance (car insurance = Event A, home insurance = Event B, overlap = people with both); bank loan applicants (high income = A, low debt = B, probability an applicant satisfies at least one condition); e-commerce behaviour (bought electronics = A, bought clothes = B, using transactional data to estimate union and overlap).
Common Mistakes: Forgetting to subtract P(A ∩ B) in non-mutually-exclusive cases; misinterpreting event relationships (treating overlapping events as if they were mutually exclusive).
Quiz Questions: - Q1: "For mutually exclusive events A and B:" → Correct Answer: "P(A ∪ B) = P(A) + P(B)" - Q2: "General addition rule is:" → Correct Answer: "P(A ∪ B) = P(A) + P(B) − P(A ∩ B)"
The General Addition Rule always works (P(A∪B) = P(A)+P(B)−P(A∩B)); the simplified "just add them" version is a special case that only holds when A and B are mutually exclusive (P(A∩B) = 0). When in doubt, use the general formula — subtracting 0 for mutually exclusive events doesn't hurt, but forgetting to subtract a genuine overlap gives a wrong (inflated) answer.
Technical note: this lesson exposed a genuine Browser-pane rendering/scroll quirk (not a course bug) — the SCORM iframe's own layout viewport (clientHeight ~666-876px) can be taller than what the outer browser pane visibly renders/allows scrolling to (~395-400px), leaving buttons and quiz content technically present in the DOM (verified via getBoundingClientRect) but outside the clickable screenshot coordinate frame even at .page-wrap's own max scroll. Fix: call resize_window with a larger explicit height (e.g. 800x900) to expose the full content, or scroll the outer top-level document too (window.scrollTo) in addition to .page-wrap. Reset the viewport to desktop preset afterward. See PROGRESS.md for full diagnostic detail — this is now documented in root context.md §8 as well.
Lesson 7: 2.1.5_Multiplication Rule of Probability
Reading Content (SLM Pages 48-49):
Let us assume that a process may be divided into two steps. If the first task can be completed in n₁ ways and the second task can be completed in n₂ ways after the first task, the method can then be carried out in (n₁ × n₂) ways. More generally, for r experiments where the first can occur in n₁ ways, the second in n₂ ways after it, the third in n₃ ways, and so on, the total possible outcomes = n₁ × n₂ × n₃ × ... × nᵣ.
Multiplicative Rule is stated as:
If A and B are two independent events, the probability of occurrence of A and B is given by: P(A ∩ B) = P(A) × P(B)
When the logical AND is used to indicate successive experiments, the "Product Rule" is applicable (as opposed to the "Sum Rule" for OR/addition).
Independent vs Dependent Events: - Independent events: the occurrence of one does not influence the occurrence of the other. Example: tossing a coin and rolling a die — P(A∩B) = P(A) × P(B). - Dependent events: the outcome/occurrence of the first affects the probability of the second. Example: drawing two cards without replacement from a deck — drawing an Ace first affects the probability of drawing a second Ace. Formula: P(A∩B) = P(A) × P(B|A), where P(B|A) = probability of B occurring given A has already occurred.
Worked Example 1: How many outcomes are there if we toss a coin and then throw a die? Answer: 2 × 6 = 12
Worked Example 2: 80% of all tourists who visit India visit Delhi (P(D)=0.8), 70% visit Mumbai (P(M)=0.7), and 60% visit both (P(D∩M)=0.6). 1. Probability a tourist visits at least one city: P(D∪M) = P(D) + P(M) − P(D∩M) = 0.8 + 0.7 − 0.6 = 0.9 2. Probability a tourist visits neither city: P(D꜀∩M꜀) = 1 − P(D∪M) = 1 − 0.9 = 0.1
Industry Use Case (Retail): Probability a customer enters a store = 0.3; if they enter, the probability they make a purchase = 0.5 (dependent events). P(Enter ∩ Buy) = 0.3 × 0.5 = 0.15 → 15% chance a passerby becomes a paying customer, useful for marketing ROI analysis.
Quiz Questions: - Q1: "For independent events A and B:" → Correct Answer: "P(A ∩ B) = P(A) × P(B)" - Q2: "If A and B are not independent:" → Correct Answer: "P(A ∩ B) = P(A|B) × P(B)"
The Multiplication Rule has two forms depending on independence: Independent events → P(A∩B) = P(A)×P(B). Dependent events → P(A∩B) = P(A)×P(B|A), using the conditional probability of B given A. Always check independence before choosing the formula — this distinction is the crux of this lesson and a natural segue into the next lesson (Conditional, Joint and Marginal Probability).
Lesson 8: 2.1.6_Conditional, Joint and Marginal Probability
Reading Content (SLM Pages 49-50):
All of the characteristics and axioms of probability apply to conditional probability. From now on, (E ∩ F) is written as EF for brevity. - The likelihood that an event will occur in light of the occurrence of another event is known as conditional probability. - The conditional probability of A given B, denoted P(A/B) or P(A|B), means "the probability of A given that B has already occurred."
Derivation (via labeled-graphic markers): 1. Probability is a measure of uncertainty that is dependent on the available data. 2. The likelihood of event E occurring may differ from the initial probability of E when we were unaware that event F would occur. 3. Conditional probability is represented as P(E|F), the likelihood that E will occur provided that F has occurred. 4. Our sample space is reduced to the event space of F if event F takes place. 5. Additionally, for event E to happen, events F and E must happen at the same time (i.e., EF must occur). 6. Therefore, the likelihood that E will occur provided that F has occurred equals the likelihood of EF (E∩F) in relation to the likelihood of F.
Formula: $$P(E|F) = \frac{P(EF)}{P(F)}$$
Another variation of the conditional probability rule (rearranged): $$P(EF) = P(E|F) \times P(F)$$
Joint, Marginal, Conditional Probability — Definitions:
| Type | Definition | Example |
|---|---|---|
| Joint Probability | P(A∩B) — probability that both A and B occur simultaneously | Probability a customer is male AND buys a product |
| Marginal Probability | P(A) — probability of a single event, irrespective of other events (derived by "marginalizing"/summing across rows or columns of a contingency table) | Probability a customer is female, regardless of purchase behavior |
| Conditional Probability | P(A | B) — probability of event A given event B has already occurred (isolates the subset of the sample space where B is true) |
Worked Example (from video, contingency table of Gender × Purchase): - Joint P(Male ∩ Buy) = 30/100 = 0.30 - Marginal P(Female) = 50/100 = 0.50 - Conditional P(Buy | Female) = 40/50 = 0.80
Worked Example (SLM, competitor product launch): If a competitor does not introduce a similar product, there's a 0.67 chance a new product succeeds; if the competitor does launch, the success chance is 0.42; there's a 35% chance the competitor launches (P(L)=0.35). - Let S = successful launch, L = competitor launches, Lᶜ = competitor doesn't launch. - P(Lᶜ) = 1 − P(L) = 1 − 0.35 = 0.65 - Using the Law of Total Probability: P(S) = P(S|L)·P(L) + P(S|Lᶜ)·P(Lᶜ) = 0.42×0.35 + 0.67×0.65 = 0.5825
Quiz Questions: - Q1: "Conditional probability P(A|B) is:" → Correct Answer: "P(A occurs given B has occurred)" - Q2: "Marginal probability refers to:" → Correct Answer: "Individual probability of an event regardless of others"
Quick mnemonic: Joint = "AND" (both together), Marginal = "alone" (ignore the other variable — the row/column total in a contingency table), Conditional = "GIVEN" (restrict the sample space to where the condition is true first).
Lesson 9: 2.1.7_Bayes' Theorem
Reading Content (SLM Pages 50-51):
For any two events E and F, we can always state that the probability of E is equal to the sum of the probability of (E ∩ F) and the probability of (E ∩ Fᶜ):
P(E) = P(E ∩ F) + P(E ∩ Fᶜ)
Derivation of Bayes' Formula: For any element in E, it must be either in both E and F, or in E but not in F. (E∩F) and (E∩Fᶜ) are mutually exclusive, since the former must be in F and the latter must not be in F. By Axiom 3 (additivity for mutually exclusive events):
P(E) = P(E∩F) + P(E∩Fᶜ) = P(E|F)×P(F) + P(E|Fᶜ)×P(Fᶜ) = P(E|F)×P(F) + P(E|Fᶜ)×[1 − P(F)]
Suppose now that E has occurred, and we want the probability that F (or one of several Fᵢ hypotheses) has occurred — using the above equations gives Bayes' formula. A market-survey framing: if events are potential "hypotheses" about some subject matter (e.g., market shares of competitors), Bayes' formula shows how those hypotheses should be adjusted in light of new experimental evidence.
Worked Example (bulbs in a bin): Three types of bulbs are in a bin. Type 1 lamps have a 0.7 probability of lasting more than 100 hours; type 2, a 0.4 probability; type 3, a 0.3 probability. The bin contains 20% type 1, 30% type 2, 50% type 3 lamps [P(T₁)=0.2, P(T₂)=0.3, P(T₃)=0.5]. Questions: (a) What is the probability a randomly chosen lamp lasts more than 100 hours? (b) Given a lamp lasted more than 100 hours, what's the probability it's type 1 / type 2 / type 3?
Solution — let S = "lasts more than 100 hours". Using the Law of Total Probability:
P(S) = P(S|T₁)P(T₁) + P(S|T₂)P(T₂) + P(S|T₃)P(T₃) = 0.7×0.2 + 0.4×0.3 + 0.3×0.5 = 0.41
Now, using Bayes' formula: - P(T₁|S) = P(S|T₁)P(T₁) / P(S) = (0.7×0.2)/0.41 = 0.341 - P(T₂|S) = P(S|T₂)P(T₂) / P(S) = (0.4×0.3)/0.41 = 0.293 - P(T₃|S) = P(S|T₃)P(T₃) / P(S) = (0.3×0.5)/0.41 = 0.366
(Check: 0.341+0.293+0.366 = 1.00 ✓ — the revised/"posterior" probabilities of each type, given the lamp survived past 100 hours, still sum to 1.)
Quiz Questions: - Q1: "Bayes' theorem is used to:" → Correct Answer: "Reverse conditional probabilities" - Q2: "In Bayes' theorem, prior probability is:" → Correct Answer: "Probability before new data"
Bayes' Theorem is the tool for "reversing" conditional probabilities: given P(E|F) (or P(S|Tᵢ) for several hypotheses), it lets you compute P(F|E) (or P(Tᵢ|S)) — i.e., go from "probability of evidence given hypothesis" to "probability of hypothesis given evidence." The prior is P(Tᵢ) (belief before new data/evidence); the posterior is P(Tᵢ|S) (updated belief after the evidence). The denominator P(S) is computed via the Law of Total Probability, summing P(S|Tᵢ)×P(Tᵢ) across all mutually exclusive, collectively exhaustive hypotheses.
Lesson 10: 2.2_Probability Distribution
Reading Content (SLM Page 52, 20-second read, no video/quiz):
- Numerous factors or features can characterise a random process; numerical values or categories describe the results — these are called random variables or random attributes.
- A list of potential values for a random variable (or possible categories for a random attribute), along with their corresponding probabilities, is called a probability distribution.
Lesson 11: 2.2.1_Introduction of Random Variables
Reading Content (SLM Page 52):
- In many real-world scenarios, a random variable exhibits a particular pattern.
- When classifying random variables, the probability mass function (PMF) is used for discrete variables and the probability density function (PDF) for continuous variables.
- All statistical computations are possible when the distribution is fully understood — but in practice, the true distribution may not be entirely known.
- By examining the mechanism that makes a variable random, it's often possible to approximate it to one of the known standard random variables/distributions — these standard distributions are also called "probability models." Once approximated to a known model, distributional properties (mean, variance, moments, etc.) can be computed via known closed-form equations.
Theoretical Distributions (4 flashcards): 1. Bernoulli Distribution 2. Binomial Distribution 3. Poisson Distribution 4. Standard deviation (flashcard content as presented by the course — likely intends "Normal Distribution" as the 4th standard theoretical distribution; captured verbatim as the platform presented it, flagged as a probable course content quirk)
Quiz Questions: - Q1: "A random variable is:" → Correct Answer: "A variable that assumes numerical values based on outcomes" - Q2: "Random variables are classified into:" → Correct Answer: "Discrete and continuous"
The 4th flashcard in this lesson's "Theoretical Distributions" set is labeled "Standard deviation" rather than a distribution name — captured as-is from the live course since it's what the platform displays, but this looks like a likely content error (standard deviation is a statistic, not a distribution; the 4th classical theoretical distribution is typically the Normal/Gaussian distribution). Worth cross-checking against the Study Material PDF if this becomes relevant to an assessment question.
Lesson 12: 2.2.2_Mean and Expected Value of Random Variable
Reading Content (SLM Pages 52-53):
- The arithmetic mean of several independent realisations of a random variable is what is intuitively understood to be the expected value of that variable in probability theory.
- It is a generalisation of the weighted average.
- The terms expectation, mathematical expectation, mean, average, and initial moment are other names for the expected value.
4-marker labeled-graphic widget ("How is random variable defined?"): 1. A collection of potential values from a random experiment is called a random variable. 2. A discrete random variable X's mean is the weighted average of all possible values that it might have. 3. The mean of a random variable assigns weight to each outcome xᵢ based on its probability, pᵢ, as opposed to the sample mean of a series of observations, which assigns equal weight to each observation. 4. The expected value of X, or the mean, is commonly represented by the symbol μ.
Formula: For discrete random variable X with values xᵢ and probabilities pᵢ: $$E[X] = \sum x_i p_i$$
Worked Example (fair coin toss game, win $1 heads / lose $1 tails): E[X] = 0.5×(1) + 0.5×(−1) = 0. A fair game has an expected value of 0 — neither player is expected to gain or lose money on average.
Quiz Questions: - Q1: "What is the expected value of a fair coin toss game where you win $1 for heads and lose $1 for tails?" → Correct Answer: "0" - Q2: "For discrete random variable X with values xᵢ and probabilities pᵢ, E[X] =" → Correct Answer: "∑xᵢpᵢ"
"Expected value" is just a weighted average where the weights are the probabilities, not equal shares — this is the key difference from a plain arithmetic/sample mean. A "fair game" is defined precisely as one where E[X] = 0.
Lesson 13: 2.2.3_Variance and Standard Deviation of Random Variable
Reading Content (SLM Page 53):
The variance is a numerical description of the spread, or dispersion, of the random variable — i.e., the variance of a random variable X is a measure of how spread out the values of X are, given how likely each value is to be observed.
Formula: $$Var(X) = \sum x^2 p - \mu^2$$
3-step process widget ("To calculate the variance"): 1. Step 1: Square each value and multiply by its probability. 2. Step 2: Sum them up and we get Σx²p. 3. Step 3: Subtract the square of the Expected Value, μ².
Standard deviation is the square root of the variance: $$\sigma = \sqrt{Var(X)}$$
Video note (animated-infographic style, 3:57): the video walks through 3 industry use cases as it plays — Stock Market Variance (daily returns spread), Quality Control (factory part-weight consistency), and a third closing segment comparing variance vs. standard deviation units — reinforcing that standard deviation is in the same units as the original data (unlike variance, which is in squared units), making it more directly interpretable.
Quiz Questions: - Q1: "If the variance of a dataset is 36, what is the standard deviation?" → Correct Answer: "6" (√36 = 6) - Q2: "Standard deviation is the:" → Correct Answer: "Square root of variance"
Variance is in squared units of the original data (hard to interpret directly); standard deviation (√variance) converts back to the original units, which is why it's the more commonly reported/interpreted measure of spread in practice.
Technical note: this lesson's video appeared to "go blank" mid-scroll at one point — a genuine transient render/interaction issue on the first attempt of this session where clicking the play button silently did nothing (console showed the same known player.php:120:19 "Cannot read properties of null (reading 'style')" error documented elsewhere in this project). Fix: re-navigated the tab to the module's launch URL (mod/scorm/view.php?id=359338) and re-entered via "Click here to start learning" — resumed at the identical bookmark and the video played normally on the next click. No special technique needed beyond the standard re-navigate fix.
Lesson 14: 2.2.4_Binomial Distribution Basics
Reading Content (SLM Pages 53-54):
- We typically carry out a large number of identical, independent trials — Bernoulli trials, each with two possible outcomes and a probability of success (p).
- Each trial yields a probability of failure (1 − p) and a probability of success (p). A random variable is binomial if it has parameters (n, p) and indicates the number of successes that occur in n trials (order of successes is not significant).
- The Bernoulli random variable has parameter (1, p), meaning n = 1 — it is a special case of the Binomial random variable.
Formula — Binomial Probability Mass Function: $$P(X = i) = \binom{n}{i} p^i (1-p)^{n-i} \quad \text{for } i = 0,1,2,...,n$$
Expected value and variance for a Binomial random variable: $$\mu = E[X] = np \qquad Var[X] = np(1-p)$$
Video note (animated-infographic style, 4:56): covers the 4 key conditions for binomial distribution (fixed number of trials, each trial independent, each trial has only two outcomes success/failure, probability of success p constant across all trials), walks through the PMF formula and its components, and a worked example.
Quiz Questions: - Q1: "Which of these is NOT a condition for binomial distribution?" → Correct Answer: "Multiple outcomes per trial" (binomial requires exactly 2 outcomes per trial — success or failure) - Q2: "What is the mean of a binomial distribution with n=10, p=0.3?" → Correct Answer: "3" (μ = np = 10 × 0.3 = 3)
The 4 conditions for a Binomial distribution (all must hold): (1) a fixed number of trials (n), (2) each trial is independent, (3) each trial has only two possible outcomes (success/failure), (4) the probability of success p is constant across all trials. If any condition is violated — especially "only two outcomes" — the binomial model doesn't apply.
Lesson 15: 2.2.5_Binomial Distribution Uses
Reading Content (SLM Pages 54-55):
Making the choice of when to employ a binomial distribution is crucial. When the following criteria are met, the binomial distribution can be used:
- Trials are repeated 'n' times and are finite (i.e., not extremely huge).
- Every trial, or random experiment, ought to be a Bernoulli trial, meaning that it should be the one that ends in success or failure.
- The chance of success in every trial is always 'p,' meaning that it never changes.
- Every trial is conducted independently.
These trials are often "with replacement" selection experiments. Taking a tiny sample from a very large population does not appreciably alter the probability of success in certain situations — therefore, the distribution could still be regarded as a Binomial (Bernoulli-trial) distribution.
Few instances of binomial distribution applications in the real world: 1. The quantity of broken light bulbs in many machine-produced goods. 2. The proportion of female births in a hospital out of n births. 3. The quantity of right responses on a multiple-choice exam. 4. The quantity of seeds in a row that sprouted after being sown. 5. The quantity of fish recovered in a sample of fish. 6. The quantity of missiles that, out of those fired, hit the targets.
3 worked-example accordions ("Click on the '+' icon to learn more about the followings"):
- Quality Control in Manufacturing: A factory produces light bulbs, each with a 95% chance of being defect-free. If a QC inspector randomly selects 20 bulbs, the binomial distribution finds the probability that exactly 18 are defect-free. Parameters: n=20, p=0.95, k=18.
- Marketing Campaign Effectiveness: An email campaign has a known 10% purchase conversion rate. If 100 emails are sent, the binomial distribution finds the probability that exactly 15 recipients purchase. Parameters: n=100, p=0.10, k=15.
- Drug Testing: A new drug is effective 70% of the time per prior studies. In a trial of 50 patients, the binomial distribution finds the probability that at least 35 patients respond positively. Parameters: n=50, p=0.70, k=35.
Quiz Questions: - Q1: "The binomial distribution is most appropriate when:" → Correct Answer: "Number of trials is fixed" - Q2: "In a batch with 5% defect rate, the chance of 2 defects in 50 items is modeled by:" → Correct Answer: "Binomial distribution"
All 3 real-world accordion examples share the same pattern: a large-ish fixed n, a known constant p, and a specific target count k — the classic signature of "use the Binomial PMF" word problems (vs. Poisson, which is used for rare-event rates over continuous time/space rather than a fixed trial count).
Lesson 16: 2.2.6_Poisson Distribution Overview
Reading Content (SLM Pages 55-56):
If, for some λ > 0, a random variable X with one of the values 0, 1, 2, ... is a Poisson random variable with parameter λ, then P(X = i) is a probability mass function (p.m.f.) of the Poisson random variable. Its expected value and variance are:
$$\mu = E[X] = \lambda \qquad Var(X) = \lambda$$
Applications for the Poisson random variable are numerous. If n is high enough and p is small enough to make the product np of reasonable size, the Poisson distribution can also be used as an approximation for a Binomial random variable with parameters (n, p) — here, np = λ is referred to as the average rate.
Typical instances where the Poisson distribution applies: 1. Daily count of accidents on the motorway. 2. The total number of earthquakes in a certain period of time. 3. The quantity of typos on a page. 4. The quantity of calls that arrive at the phone exchange each minute. 5. A server's interrupt rate per second.
Video content (4:35, Dr. Deepali Salwan): - What is Poisson Distribution? Models the number of times an event occurs in a fixed interval of time or space; events happen independently and at a constant average rate λ; useful for rare or random events. - Poisson Probability Formula: $$P(X = k) = \frac{\lambda^k e^{-\lambda}}{k!}$$ where k = number of events, λ = average rate of occurrence, e = Euler's number (≈2.71828). - Worked Example: A call center receives 5 calls per hour on average (λ=5). What's the probability exactly 3 calls arrive in the next hour? $$P(X=3) = \frac{5^3 e^{-5}}{3!} = \frac{125 \times 0.0067}{6} \approx 0.1404$$
Quiz Questions: - Q1: "λ in Poisson distribution means:" → Correct Answer: "Average rate of events" - Q2: "Poisson distribution assumes:" → Correct Answer: "Constant average event rate"
Poisson vs. Binomial — the recurring exam distinction: Binomial counts successes over a fixed number of discrete trials (n, p known); Poisson counts events over a continuous interval (time/space) at a constant average rate λ, with no natural "n" or "p" — only λ. Poisson can also approximate Binomial when n is large and p is small (np = λ stays moderate). Uniquely for Poisson, mean = variance = λ.
Lesson 17: 2.2.7_Poisson Distribution Applications
Reading Content (SLM Pages 56-57):
The Poisson probability mass function (p.m.f.) for a given λ can be easily calculated using scientific calculators. However, calculating cumulative probabilities, i.e., the cumulative distribution function (c.d.f.), manually can be quite tedious. In such cases, we can utilise cumulative Poisson probability tables or software tools that provide cumulative probabilities directly.
Finding cumulative Poisson probability from tables: 1. Look at the given value of λ (average rate) in the first column of the table. 2. In the first row, look for the value of i (the number of successes). 3. Locate the cell at the intersection of the i-column and λ-row — this cell contains the cumulative Poisson probability.
Video content (3:32, Dr. Deepali Salwan) — "Poisson Distribution Applications": - Recap of Poisson Distribution: Models count of independent events over fixed intervals; parameter λ = average number of events. - Application in Healthcare: Modelling patient arrivals in emergency rooms; resource allocation based on expected arrivals. - Application in Retail: Customer arrivals or sales transactions in fixed periods; guides staffing and inventory planning. - Case Study — Call Center Staffing: Average 20 calls/hour (λ=20). Calculate probability of more than 25 calls in an hour: P(X>25) = 1 − P(X≤25), computed via software/tables to decide if extra agents are needed. - Benefits of Poisson Modeling: Efficient for rare event modelling; enables predictive analytics; simple to calculate and interpret.
Worked Example (Poisson as Binomial approximation, via "Method I"/"Method II" accordion): A scenario with n=10, p=0.1 (per-item defect-type probability), asking for P{X<1} = P(0)+P(1). - Method I (exact Binomial): P{X<1} = ₁₀C₀(0.1)⁰(0.9)¹⁰ + ₁₀C₁(0.1)¹(0.9)⁹ = 0.7361 (also readable directly off a Cumulative Binomial Probabilities table for n=10, p=0.1, i=1). - Method II (Poisson approximation): λ = np = 10×0.1 = 1. P{X<1} = e⁻¹(1)⁰/0! + e⁻¹(1)¹/1! = e⁻¹ + e⁻¹ = 0.7358 (also readable off a Cumulative Poisson Probabilities table for λ=1, i=1). - Note: the Poisson approximation (0.7358) is very close to the exact Binomial value (0.7361) — confirming Poisson gives a reasonably good approximation to Binomial when n is large and p is small.
Second Worked Example (bank clerk passbook updates): Average time to update a passbook is 15 seconds. Someone arrives just ahead of you. Find the probability you'll wait (1) more than 1 minute, (2) less than ½ minute. - λ = 60/15 = 4 passbooks per minute (rate per minute, since 60 seconds ÷ 15 seconds/passbook). - P{X>1} = 1 − F(1) = e⁻⁴ = 0.0183 - P{X<0.5} = F(0.5) = 1 − e⁻² = 1 − 0.1353 = 0.8647
Quiz Questions: - Q1: "Which industry commonly uses Poisson modeling?" → Correct Answer: "Call centers" - Q2: "Poisson distribution best models:" → Correct Answer: "Rare events over fixed intervals"
The bank-clerk example is a good illustration of converting units to match the interval: since the average service time is given as 15 seconds/passbook, converting to a per-minute rate (λ = 60/15 = 4) is the necessary first step before applying the Poisson formulas — always check that λ and the time window in the question are expressed in matching units.
Lesson 18: 2.2.8_Normal Distribution and Empirical Rule
Reading Content (SLM Pages 57-59):
Many commercial and engineering challenges involve the normal random variable and its distribution. Under some circumstances, a variety of other distributions — including the binomial, Poisson, beta, chi-square, student's, exponential, etc. — could also be roughly approximated by the normal distribution (usually in cases with huge sample sizes).
Random variables that closely resemble the normal distribution are those that are affected by a large number of independent causes, where the impact of each individual cause is not statistically significant when compared to the combined effect of all others.
Examples of normally distributed random variables: 1. Weights of coffee filled in packs 2. Lengths of nails manufactured on a machine 3. The hardness of ball bearing surfaces 4. Diameters of shafts produced on a lathe, and the effectiveness of training programmes on employee productivity
Additionally, a large number of sampling statistics have a normal distribution, such as sample means (X̄).
Video content (4:54, Dr. Deepali Salwan): - Recap: Continuous, symmetric, bell-shaped distribution; mean, median, and mode are equal and at the centre; total area under the curve is 1 (100%); mean (μ) = centre of the distribution, standard deviation (σ) = measures spread. - Standard Normal Distribution: Mean μ=0, Standard Deviation σ=1. Introduction to Z-scores for standardizing normal distribution data. Formula for Z-score: $Z = \frac{x - \mu}{\sigma}$ - Why Normal Distribution matters: Models many natural and social phenomena (heights, test scores, etc.); foundation for many statistical methods and probability calculations; predicts the likelihood of outcomes in a population. - When to use Normal Distribution: (1) When data is continuous and symmetrically distributed; (2) when mean and standard deviation are known; (3) common in quality control, finance, research, and social sciences. - Empirical Rule (68-95-99.7 Rule): Also called the Three-Sigma Rule. For bell-shaped data: 68% within 1σ of mean (μ), 95% within 2σ, 99.7% within 3σ. - Use Case Example: Lifespan of animals in a zoo is normally distributed with mean = 13.1 years, σ = 1.5 years. 68% live between 11.6 and 14.6 years (μ±1σ); 95% live between 10.1 and 16.1 years (μ±2σ); 99.7% live between 8.6 and 17.6 years (μ±3σ).
Empirical Rule — 3 components: 1. 68% of the data falls within the first standard deviation from the mean. 2. 95% fall within two standard deviations. 3. 99.7% fall within three standard deviations.
The empirical rule is often used in statistics for forecasting, especially when obtaining the right data is difficult or impossible to get — it gives a rough estimate of what your data collection might look like if you were able to survey the entire population.
Normal Distribution Probability Density Function (p.d.f.): for a random variable X with parameters μ and σ: $$f(x) = \frac{1}{\sigma\sqrt{2\pi}} e^{-\frac{(x-\mu)^2}{2\sigma^2}}, \quad -\infty < X < \infty$$
Properties of the p.d.f.: - This distribution is symmetric around μ and has a bell-shaped curve. - It provides a theoretical foundation for the finding that many random events in real life roughly follow a normal probability distribution. - The variance of a normal random variable is Var(X) = σ², and its mean is E(X) = μ. - Another random variable with parameters (aμ+b) and (aσ) is also normally distributed if X is normally distributed with parameters μ and σ (linear transformation property).
Properties of Normal Distribution (6 points): 1. About the mean μ, it is fully symmetric. 2. The mean, median and mode of a normal distribution are equal. 3. It has a single mode, skewness = 0 and kurtosis = 0. It is unimodal. 4. Standard deviation is a limiting form of the binomial distribution in cases where the probability p and (1−p) are not extremely tiny and the number of trials n is considerable. 5. When mean μ = λ is very big, the normal distribution is a limiting case of the Poisson distribution. 6. We typically utilise normal distribution (or, more frequently, standard normal distribution) tables when working on the probability of a normal distribution.
While reading these tables, properties are: - The probability that a normally distributed random variable with mean μ and variance σ² lies between two specified values a and b is P(a < X < b) = area under the curve P(x) between the specified values X=a and X=b. - The total area under the curve P(x) is equal to 1, in which 0.5 lies on either side of the mean.
Quiz Questions: - Q1: "Which statement about the Empirical Rule is correct?" → Correct Answer: "99.7% of data falls within 3 standard deviations of the mean" - Q2: "A normal distribution is:" → Correct Answer: "Symmetric bell-shaped curve"
The Empirical Rule (68-95-99.7 Rule / Three-Sigma Rule) is a cornerstone fact for any normally distributed variable: 68% of data falls within μ±1σ, 95% within μ±2σ, and 99.7% within μ±3σ. This is the fastest way to estimate spread/probability for bell-shaped data without needing full Z-tables, and is a near-certain assessment topic.
Technical note: this lesson hit the known viewport-height scroll bug (SCORM iframe content taller than the Browser pane's visible capture area) partway through the reading content — fixed via resize_window to 800×900 as documented in context.md §8, then reset to desktop preset after completing the lesson.
Lesson 19: 2.2.9_Standard Normal Distribution Intro
Reading Content (SLM Pages 59-60):
Integration is required to calculate the cumulative density of a normal distribution. Moreover, another issue with tabulation is that it is impractical to generate tables for every potential value of μ and σ². As a result, we convert the normal random variable to the standard normal random variable, using a transformation:
$$Z = \frac{(x-\mu)}{\sigma} = \frac{1}{\sigma}x - \frac{\mu}{\sigma}$$
It is a normally distributed random variable with parameters μ = 0 and σ = 1. Any normal random variable can be transformed into the standard normal random variable Z. The cumulative distribution function:
$$F(a) = \int_{-\infty}^{a} f(x)\,dx = \int_{\infty}^{a} \frac{1}{\sqrt{\sqrt{2\pi}}} e^{-\frac{z^2}{2}}\,dz$$
This has been calculated for various values of 'a' and tabulated. Also: F(−a) = 1 − F(a) and F(a < Z < b) = F(b) − F(a).
Video content (~6:39, Dr. Deepali Salwan) — "Standard Normal Distribution Intro": - What is Standard Normal Distribution — Key Characteristics: Symmetric (bell-shaped curve centered at zero); Mean (μ)=0 (the centre point); Standard Deviation (σ)=1 (measures spread); Total Area=1 (represents 100% probability); Asymptotic (curve never touches the x-axis). Notation: Z ~ N(0, 1). - How Does It Work — Step-by-Step Process: (1) Identify the original value (X), mean (μ), and standard deviation (σ); (2) Subtract the mean from the original value (X−μ); (3) Divide the result by the standard deviation; (4) The result is your Z-score; (5) Use Z-table to find probabilities. Formula: Z = (X − μ) / σ. - When to Use Standard Normal Distribution — Perfect Scenarios: Data follows normal distribution; need to compare different datasets; calculating probabilities; hypothesis testing; quality control processes; standardizing test scores. - Key Indicators: Bell-shaped data distribution; large sample size (n ≥ 30); Central Limit Theorem applies; need for probability calculations; comparing across different scales; statistical inference required. - Real-World Use Case — Healthcare (Patient Vital Signs): Blood pressure readings are standardized to identify patients outside normal ranges — helps doctors quickly assess risk levels across different patient populations. - Key Properties & Rules: Symmetry: P(Z>a) = P(Z<−a); Total Probability: P(−∞<Z<∞) = 1; Median = Mode = Mean = 0; Inflection Points at Z=±1; Standard Deviation = 1. Worked values: P(Z≤0)=0.5, P(Z≤1.96)≈0.975. - Empirical Rule (68-95-99.7 Rule) restated in Z-terms: 68% of data falls within 1σ (Z=±1); 95% within 2σ (Z=±2); 99.7% within 3σ (Z=±3). - Practical Applications — Business Applications: Sales forecasting; inventory management; customer behaviour analysis; risk management; performance benchmarking. - Worked Example (video): Given heights with μ=170cm, σ=10cm, find P(height > 185cm). Z = (185−170)/10 = 1.5. P(Z>1.5) = 1 − 0.9332 = 0.0668 or 6.68%.
Worked Example (SLM reading, tea-packing machine — 99% Confidence Interval): A machine fills 200gm tea bags with a 0.25gm² variance. Customers would refuse (and it would be illegal) to accept packs under 200gm, so the production manager is asked to programme the machine to slightly overfill — but Finance objects since overfilling causes financial loss. The general manager needs the 99% confidence interval when the machine is set at 200gm.
Solution: Let X = weight of tea in a pack (random variable). μ = 200gm, variance σ² = 0.25gm² → σ = 0.5gm. - For 99% confidence, half the area under the (symmetric) curve = 0.99/2 = 0.495. - The Z-value corresponding to probability 0.495 is z = 2.575. - 99% CI in terms of Z: ±2.575. In terms of X: x = σz + μ = 0.5×(±2.575) + 200 = 200 ± 1.2875, i.e. (198.71 to 201.29). - Recommendation: programme the machine to fill tea with a mean weight of 201.2875 (≈201.29)g — this gives 99.9% certainty of complying with regulatory requirements (packs ≥200gm) while minimizing the cost of overfilling.
Quiz Questions: - Q1: "What are the mean and standard deviation of a standard normal distribution?" → Correct Answer: "μ = 0, σ = 1" - Q2: "In standard normal distribution, mean is:" → Correct Answer: "0"
The Standard Normal Distribution is simply the normal distribution with μ=0 and σ=1 — every normal random variable X can be converted to it via the Z-score transformation Z=(X−μ)/σ, which is the single most useful trick in this module: it lets a single Z-table serve every possible normal distribution instead of needing infinite tables for every μ,σ combination.
Technical gotcha (new this lesson): the browser tab's real-mouse scroll can silently "freeze" mid-lesson (repeated scroll actions return an identical screenshot even though .page-wrap.scrollTop genuinely isn't at max) — the fix that worked was a real left_click directly on the lesson content area first (to (re)focus the scrollable container), after which scroll resumed working normally. Distinct from the documented "stuck tab" bug (context.md §8) since no tab re-navigation was needed — just a focus click.
Lesson 20: 2.2.10_Applications of Standard Normal Distribution
Reading Content (SLM Pages 60-61):
In many different domains, the normal distribution is used to simulate the behavior of random variables.
Uses of Normal Distribution: 1. Finance — simulating the behaviour of interest rates, stock prices, and other financial variables. 2. Engineering — modelling the distribution of measurement errors, strength of materials, and other engineering factors. 3. Social sciences — simulating the distribution of human attributes such as height, weight, and IQ scores. 4. Natural sciences — simulating how observations of physical quantities like mass, pressure, and temperature are distributed. 5. Statistics — many statistical tests and processes (regression analysis, confidence intervals, hypothesis testing) are based on the normal distribution.
Video content (6:22, Dr. Deepali Salwan) — "Applications of Standard Normal Distribution": - Key Concepts recap (68-95-99.7 Rule): 68% within 1σ, 95% within 2σ, 99.7% within 3σ. Z-table used to find probabilities and percentiles. Critical Values = Z-scores that define specific probability regions. - Example: If a student scores 85 on a test where the mean is 75 and standard deviation is 10: Z=(85−75)/10=1.0 — this score is 1 standard deviation above the mean. - Step-by-Step Process: (1) Calculate Z-score: Z=(X−μ)/σ; (2) Use Z-table: look up the Z-score to find probability; (3) Interpret result: convert to percentage or make a decision. - Worked Example (bolts): What percentage of bolts are longer than 5.15cm (μ=5.0cm, σ=0.1cm)? Z=(5.15−5.0)/0.1=1.5. From Z-table: P(Z<1.5)=0.9332. P(Z>1.5)=1−0.9332=0.0668 or 6.68%. - History and Uses (Historical Development): 1733 — Abraham de Moivre first described the normal curve; 1809 — Carl Friedrich Gauss developed the theory; 1900s — widespread adoption in quality control and research; Today — essential tool in data science and analytics. - Use Case 1 — Medical Diagnosis (Blood Pressure Analysis): A hospital analyses blood pressure readings — mean systolic BP: 120 mmHg, standard deviation: 15 mmHg, patient reading: 150 mmHg. - Use Case 2 — Educational Assessment (Standardized Test Scoring): SAT Score Analysis — mean SAT score: 1060, standard deviation: 195, student score: 1300. Z=(1300−1060)/195=1.23. From Z-table: this student scored better than 89% of test takers. Interpretation: excellent performance, qualifies for top universities.
Quiz Questions: - Q1: "If X has a normal distribution with μ = 50 and σ = 10, what is the Z-score for X = 65?" → Correct Answer: "1.5" (Z=(65−50)/10=1.5) - Q2: "Z-scores are used to" → Correct Answer: "Compare values across distributions"
The recurring 3-step Z-score workflow across every worked example in this lesson: (1) compute Z=(X−μ)/σ, (2) look up the cumulative probability P(Z<z) in a Z-table, (3) interpret — subtract from 1 for "greater than" questions, or read the percentile directly for "better than X% of..." questions. This is the single most reusable calculation pattern in the whole probability-distributions module.
Lesson 21: 2.2.11_Relevant Industry Example
Reading Content (SLM Pages 61-65) — Illustrative Example (Climbing Helmets):
A business selling climbing gear currently has children's climbing helmets on the market and wants to expand to adult helmets. The company wants to know how many of each size (small/medium/large, for males and females separately) to manufacture. The business assumes a person's head size roughly follows a normal distribution, with different variances and means for men and women — so the owner can order the right amount and size of helmets by modeling head size with the normal distribution.
How to depict a normal distribution: A normal random variable has a bell-shaped PDF symmetric about its mean μ. The curve extends to ±infinity even though only a finite range is typically plotted. The mean (μ) controls where the curve's peak sits on the X-axis (decreasing μ shifts the curve left, increasing shifts it right); the standard deviation (σ) controls the curve's spread (higher σ = wider/flatter curve, lower σ = narrower/taller curve) — illustrated by comparing mean=0,s.d.=1 (tall/narrow) vs. mean=0,s.d.=2 (short/wide).
Compute Probabilities: For X ~ N(0,1), the shaded area P(0 ≤ X ≤ 1) = 0.3413 (from standard normal tables). Also: 95% of the area under N(μ,σ²) lies in the interval μ − 1.96σ < X ≤ μ + 1.96σ — a useful memory aid is "95% of the area is within about 2 standard deviations of the mean."
Uniformity (standardization recap): The standard normal distribution has σ=1, μ=0. To standardize a non-standard normal random variable X (mean μ≠0, σ≠1): subtract the mean, then divide by the standard deviation → the result is called Z. Z can then be used with standard normal tables to calculate probabilities.
Video content (~8:54, Dr. Deepali Salwan) — "Relevant Industry Example" (E-commerce Analytics): - Learning Objectives: Apply probability and distribution concepts to e-commerce data; use statistical models for business decision-making. - Business Context: An e-commerce company analyses customer behaviour, order patterns, and marketing effectiveness across multiple channels to optimize operations and increase profitability. Customers: 100,000 active users. Daily Orders: average 500 orders. Conversion Rate: 3% of visitors make a purchase. Return Rate: 8% of orders returned. Marketing Channels: Email, Social Media, PPC. - Statistical Applications — Descriptive Statistics: Calculate mean, median, variance for sales. Example: Store 001's sales mean over 12 months = 49.92K. - Probability Modelling – Binomial Distribution: 30% purchase rate — probability exactly 10 of 40 customers purchase: P(X=10) = C(40,10)×(0.3)¹⁰×(0.7)³⁰ ≈ 0.122. - Probability Modelling – Poisson Distribution (Order Arrivals): Average 20 orders/hour peak time — probability exactly 25 orders in next hour: P(X=25) = (20²⁵×e⁻²⁰)/25! ≈ 0.044. - Normal Distribution – Order Values: μ=$85, σ=$25. Probability order > $100: P(X>100) = P(Z>0.6) = 0.274. - Scatter Plot & Trend Line: Plot advertising spend vs. sales — positive correlation observed. Regression equation: Sales = 10 + 8 × Advertising Spend. - Outlier Detection – Box Plot: Store #57's sales = 75K (above upper quartile 60K), potential outlier. - 1. Binomial Distribution — Conversion Analysis: Email campaign to 10,000 customers, 3% conversion rate. P(exactly 300 conversions) = C(10000,300)×(0.03)³⁰⁰×(0.97)⁹⁷⁰⁰. Expected conversions: μ=np=10,000×0.03=300. Standard deviation: σ=√[np(1−p)]=√[300×0.97]=17.05. - 2. Poisson Distribution — Order Arrivals: Average 20 orders/hour during peak time. P(exactly 25 orders in next hour) = (20²⁵×e⁻²⁰)/25! ≈ 0.044. - 3. Normal Distribution — Order Values: μ=$85, σ=$25. P(order > $100) = P(Z > 0.6) = 0.274. - Bayes' Theorem Application: Customer segments: Premium (20%), Regular (80%). Purchase probability: P(Purchase|Premium)=0.15, P(Purchase|Regular)=0.02. If a customer makes a purchase, probability they're Premium: P(Premium|Purchase) = (0.15×0.20)/[(0.15×0.20)+(0.02×0.80)] = 0.652. - Business Insights: Inventory Management → use Poisson for demand forecasting. Marketing ROI → Binomial models for campaign effectiveness. Customer Segmentation → Bayesian updating for personalization. Quality Control → Normal distribution for return rate analysis. Pricing Strategy → Expected value calculations for optimal pricing.
Quiz Questions: - Q1: "In telecom, Poisson distribution is useful for:" → Correct Answer: "Modeling call arrivals" - Q2: "Binomial distribution is best for:" → Correct Answer: "Predicting repeated failures in a process"
This lesson is a capstone integration of the whole module — a single e-commerce business scenario walks through Binomial (conversion/campaign analysis), Poisson (order-arrival forecasting), Normal (order-value analysis), and Bayes' Theorem (customer segmentation) all applied to the same underlying business, reinforcing when to reach for each distribution: fixed trials with success/failure → Binomial; rare events over a time interval → Poisson; continuous, symmetric real-valued data → Normal; updating a belief given new evidence → Bayes'.
Lesson 22: 2.2.12_Case Study
Reading Content (SLM Pages 65-66) — Case Study: Manufacturing Quality Assurance (Tabs: CONTEXT / GOAL): - Context: An electronics manufacturer regularly examines the size of a component for quality control. The dimensions follow a normal distribution. - Goal: To evaluate the quality control data and decide whether to approve or reject the product using the conventional (standard) normal distribution.
Actions (3-item accordion): 1. Data collection: A sample of fifty components is gathered by the quality control team, who then measure each component's dimensions. 2. Analysing Data: The component dimensions' mean and standard deviation are computed. The standard normal distribution is expressed as N(0,1), where 0 is the mean and 1 is the standard deviation — the mean dimension is represented by μ and the standard deviation by σ. 3. Z-Score Estimate: The Z-score for each component is determined by applying the formula Z=(X−μ)/σ, where X is the measurement of each unique dimension.
Interpretation: - The Z-score shows the number of standard deviations a given measurement deviates from the mean. - Z-score > 0 → measurement is above the mean; Z-score < 0 → measurement is below the mean. - Z-scores can be used to calculate the likelihood that a measurement falls within a specific range. - Acceptance Standards: the company sets a tolerance range — Z-scores within ±2 standard deviations (−2 ≤ Z ≤ 2) are deemed acceptable; Z-scores outside this range flag components for rejection or additional examination. - Making Choices: the QC team accepts/denies each component based on its Z-score — parts within range ship/proceed further; parts outside range are reworked or discarded.
Conclusion: By using the standard normal distribution, the manufacturer can proficiently analyse quality control data, detect deviations from intended requirements, and make informed accept/reject decisions — promoting steady product quality and customer satisfaction.
Video content (~8:42, Dr. Deepali Salwan) — "Probability Theory: Case Study" (Supply Chain Risk Management): - Learning Objectives: Integrate all probability concepts in comprehensive analysis; develop risk management strategies using statistical methods; create actionable business recommendations from statistical analysis. - Statistical Analysis Framework: 1. Supplier Reliability (Binomial Distribution): Supplier A has a 95% on-time delivery rate over 100 shipments. P(≥90 on-time deliveries) = Σ C(100,k)×(0.95)ᵏ×(0.05)¹⁰⁰⁻ᵏ for k=90 to 100. 2. Demand Forecasting (Normal Distribution): Weekly demand μ=10,000 units, σ=2,500 units. Safety stock calculation: P(Demand ≤ Safety Stock + Expected)=0.95. Safety Stock = μ + 1.645σ = 10,000 + 1.645(2,500) = 14,113 units. 3. Transportation Delays (Poisson Distribution): Average 2 delays per month for a shipping route. P(≤1 delay next month) = P(X=0)+P(X=1) = e⁻²+2e⁻² = 3e⁻² ≈ 0.406. - Risk Assessment Using Bayes' Theorem — Supplier Risk Categories: High Risk P(H)=0.1, Medium Risk P(M)=0.3, Low Risk P(L)=0.6. Delay Probabilities: P(Delay|High)=0.3, P(Delay|Medium)=0.1, P(Delay|Low)=0.02. If a delay occurs, probability the supplier is high risk: P(High|Delay) = (0.3×0.1) / [(0.3×0.1)+(0.1×0.3)+(0.02×0.6)] = 0.455. - Expected Value Analysis — Cost Structure: Normal Operations: $1M/month. Minor Disruption: $1.5M/month (probability 0.1). Major Disruption: $3M/month (probability 0.02). Severe Disruption: $5M/month (probability 0.005). Expected Monthly Cost: E(Cost) = 1.0(0.885)+1.5(0.1)+3.0(0.02)+5.0(0.005) ≈ $1.14M. - Risk Mitigation Strategies — 1. Diversification Strategy: Multiple Suppliers (reduce single-point failure risk); Geographic Distribution (minimise regional disruptions); Transportation Modes (alternative delivery options). - Statistical Monitoring System — Control Charts: X-bar charts (monitor delivery times); P-charts (track defect proportions); C-charts (count customer complaints). - Business Recommendations — Long-term Strategy: (1) Build resilient supply network with multiple suppliers; (2) Invest in predictive analytics for demand forecasting; (3) Develop supplier development programs; (4) Create integrated risk management dashboard. - Financial Impact Analysis: Current Annual Risk Cost: $2.4M. Implementation Cost: $500K. After Implementation: $1.8M (25% reduction). ROI: 120% in first year.
Quiz Questions: - Q1: "A bank analyzed customer complaints over time using a Poisson distribution. What key feature of Poisson distribution justified this choice?" → Correct Answer: "It is suitable for rare events in a fixed interval" - Q2: "A manufacturing company uses the binomial distribution to predict defective items in a batch. Which assumption must hold true for the binomial model to be valid?" → Correct Answer: "The probability of defect is constant and trials are independent"
Safety stock formula (Normal distribution application): Safety Stock = μ + z×σ, where z is the Z-score for the desired service level (z=1.645 for 95% one-tailed confidence). This is a widely-used real-world inventory management formula built directly on Z-score theory from this module.
Technical gotcha (reinforces earlier finding): this lesson's continue button stayed locked even after the video, both quizzes, both tabs, and all 3 accordion items were genuinely completed and .page-wrap.scrollTop was confirmed at max via JS — large 10-tick scroll jumps were NOT sufficient to trigger the scroll-reveal completion tracker. Fix: redid the full top-to-bottom scroll using smaller 5-tick increments with a ~1s wait between each step (matching the documented "4 ticks + wait" protocol in context.md §19) — this immediately unlocked the button. Lesson: when scrollTop is already at max but the lock persists, don't assume it's a genuine bug — retry the full scroll with smaller/slower increments before escalating.
Module Summary (Lesson 23) and Glossary (Lesson 24)
Module Summary — key recap points from the course's own Summary screen: - Probability is the quantitative measure of risk (Statistician I.J. Good: "The theory of probability is much older than the human species, since the assessment of uncertainty incorporates the idea of learning from experience, which most creatures do.") - A probability event is a set of outcomes from an experiment — a subset of the sample space (the entire potential set of outcomes of a random experiment). - Conditional probability is the likelihood that an event will occur in light of the occurrence of another event; P(A|B) means "the probability of A given that B has already occurred." - Bayes' formula shows how "hypotheses" (e.g., market shares of competitors) should be adjusted in light of new experimental evidence. - The expected value of a random variable is intuitively the arithmetic mean of several independent realisations of that variable.
Module Glossary — key terms from the course's own Glossary screen: - Probability: An expression of the likelihood or chance of occurrence of an event — a number ranging from zero to one. - Continuous Probability Distributions: Distributions arising from continuous random variables (variables that take on any value, including fractions/decimals) — the opposite of discrete. - Random Experiment: A process or activity that results in outcomes under study (e.g., sampling from a production lot). - Sample: The part of the universe selected for investigation, exhibiting the characteristics of the universe ("sample" literally means "small universe"). - Bayes' Theorem: Named after Reverend Thomas Bayes — describes the probability of an event based on prior knowledge of conditions that might be related to the event.
Both Lessons 23 (Summary) and 24 (Glossary) are pure reference/recap screens — no video or quiz, just reveal-on-scroll bullet content followed by a continue button. Both unlocked immediately after a full top-to-bottom scroll, no widgets to interact with.
Lesson 25: Check Your Understanding — 20/20 = 100%
A 20-question self-check quiz card widget (passing threshold 0%, so it always "passes," but answered every question deliberately and correctly). Topics covered the full module: event algebra, conditional probability, random variable classification, variance, binomial distribution, sample space, classical probability calculations (dice, cards, decks), mutually exclusive events, Bayes-adjacent definitions, addition/sum rule, independence vs. dependence, and the fundamental [0,1] bound on probability. See Quiz Bank Lesson 25 section for the full Q&A list. Final score: 100% (20/20).
Lesson 26: Exercise (Self-Study, Written Questions)
Reflective/written exercises intended for the learner's notebook — not graded: 1. Explain the concept of intersection of events using the example of rolling a pair of dice (both dice showing the same number vs. sum being 6). 2. Describe the Addition Rule of Probability with a practical mutually-exclusive-events scenario. 3. Explain the criteria for employing a binomial distribution, with real-world examples where those criteria are met. 4. Provide an overview of the Poisson distribution — key characteristics, applications, and parameters. 5. Explore the normal distribution and the Empirical Rule — characteristics, mean/SD definition, and percentages within 1/2/3 standard deviations.
Lesson 27: Learning Activities (Self-Study, Written Activities)
- A multi-step science-experiment scenario (5 test tubes × 3 chemical solutions) — frame a question using the Multiplication Rule of Probability to calculate total possible outcomes, and generalize to more than two steps.
- Investigate student heights as a random variable — discuss discrete vs. continuous random variables, and classify another real-life random variable as discrete or continuous.
Lesson 28: Further Readings
- Richard I. Levin, David S. Rubin, Sanjay Rastogi, Masood Husain Siddiqui — Statistics for Management, Pearson Education, 7th Edition, 2016.
- Prem S. Mann — Introductory Statistics, 7th Edition, Wiley India, 2016.
- Gareth James, Daniela Witten, Trevor Hastie, Robert Tibshirani — An Introduction to Statistical Learning with Applications in R, Springer, 2016.
Lesson 29: Bibliography (End of Module)
- Srivastava V.K. et al. — Quantitative Techniques for Managerial Decision Making, Wiley Eastern Ltd.
- Richard I. Levin and Charles A. Kirkpatrick — Quantitative Approaches to Management, McGraw Hill, Kogakusha Ltd.
- Prem S. Mann — Introductory Statistics, 7th Edition, Wiley India, 2016.
- Budnick, Frank S., Dennis McLeavey, Richard Mojena — Principles of Operations Research, AITBS, New Delhi.
- Sharma J.K. — Operations Research: Theory and Applications, MacMillan, New Delhi.
Module 2 (Probability and Probability Distributions) — all 29 lessons COMPLETE, 2026-09-25. Covered: Probability Theory fundamentals, Types/Algebra of Events, Addition & Multiplication Rules, Conditional/Joint/Marginal Probability, Bayes' Theorem, Random Variables (discrete/continuous), Expected Value, Variance/SD, Binomial Distribution, Poisson Distribution, Normal Distribution, Standard Normal Distribution & Z-scores, a comprehensive Relevant Industry Example (e-commerce analytics), and a Case Study (manufacturing QA + supply chain risk management). Self-check quiz scored 100%. Next: Module 2 Assessment.
✨ Quick Revision Pack (Module 2)
60-second summary
- Probability P(A) = favourable / total outcomes, between 0 and 1. Complement: P(A') = 1 - P(A).
- Addition rule: P(A ∪ B) = P(A) + P(B) - P(A ∩ B) (mutually exclusive: no overlap term).
- Multiplication rule: P(A ∩ B) = P(A) x P(B | A); independent events: P(A) x P(B).
- Conditional probability: P(A | B) = P(A ∩ B)/P(B). Bayes: P(A | B) = P(B | A)P(A) / P(B) (revise a prior with new evidence).
- Random variables: E(X) = Σ x·P(x); Var(X) = Σ (x - μ)² P(x).
- Binomial (n trials, fixed p): mean np, variance npq. Poisson (rare events per interval): mean = variance = λ. Normal: bell-shaped; empirical rule 68-95-99.7; standardise with z = (x - μ)/σ.
Diagrams
Choosing a distribution
flowchart TB
Q{"What kind of count or measure?"} -->|"Number of successes in n fixed trials, yes/no, constant p"| B["Binomial B(n, p)"]
Q -->|"Number of rare events in a fixed interval or area"| P["Poisson (λ)"]
Q -->|"Continuous measurement, symmetric bell shape"| N["Normal (μ, σ)"]
Empirical rule (normal distribution)
| Range | Share of data |
|---|---|
| μ ± 1σ | about 68% |
| μ ± 2σ | about 95% |
| μ ± 3σ | about 99.7% |
Bayes' theorem in a tree
flowchart TB S["Start"] -->|"P(A1)"| A1["Event A1"] -->|"P(B | A1)"| B1["B occurs"] S -->|"P(A2)"| A2["Event A2"] -->|"P(B | A2)"| B2["B occurs"] B1 & B2 --> R["P(A1 | B) = P(A1)P(B | A1) / [P(A1)P(B | A1) + P(A2)P(B | A2)]"]
Memory tricks
"OR → add, AND → multiply." P(A or B) adds (subtract the overlap); P(A and B) multiplies (use the conditional if dependent).
Binomial = "BINS": Binary outcomes, Independent trials, Number of trials fixed, Same probability each time.
Poisson: mean = variance = λ. If the question says "on average 3 per hour", then λ = 3 for one hour (scale λ to the interval: 2 hours → λ = 6).
z-score tells how many standard deviations a value is from the mean. P(z < 0) = 0.5; P(-1 < z < 1) = 0.6826.
Mutually exclusive ≠ independent. Exclusive events cannot happen together (so they are dependent, P(A ∩ B) = 0).
Formula sheet
| Concept | Formula |
|---|---|
| Binomial P(X = r) | nCr · pʳ · qⁿ⁻ʳ |
| Binomial mean, variance | np ; npq |
| Poisson P(X = r) | e⁻ᵏ λʳ / r! |
| Normal standardisation | z = (x - μ)/σ |
| Expected value | E(X) = Σ x P(x) |
| Variance of X | E(X²) - [E(X)]² |
| Bayes | P(A |
Practice problems (with answers)
- Addition: P(A) = 0.5, P(B) = 0.4, P(A ∩ B) = 0.2. → P(A ∪ B) = 0.5 + 0.4 - 0.2 = 0.7.
- Conditional: P(A ∩ B) = 0.2, P(B) = 0.4. → P(A | B) = 0.5.
- Independent events: P(A) = 0.3, P(B) = 0.5 → P(both) = 0.15.
- Binomial: a coin is tossed 5 times; P(exactly 2 heads). → 5C2 (0.5)⁵ = 10/32 = 0.3125.
- Poisson: 2 defects per metre on average; P(no defect in 1 metre). → e⁻² = 0.1353.
- Normal: marks μ = 60, σ = 10. P(X > 70)? → z = 1 → 1 - 0.8413 = 0.1587.
- Expected value: payoffs 10, 20, 30 with probabilities 0.2, 0.5, 0.3. → 2 + 10 + 9 = Rs 21.
- Bayes: machine A makes 60% of items (3% defective); machine B 40% (5% defective). An item is defective; P(from A)? → (0.6 x 0.03) / (0.6 x 0.03 + 0.4 x 0.05) = 0.018/0.038 = 0.474.
- Short answer (5 marks): State the properties of the normal curve. → Symmetric, bell-shaped, mean = median = mode, asymptotic to the x-axis, total area 1, 68-95-99.7 rule.