From Descriptive to Inferential: What This Course Adds
Statistical Inference & Applied Statistics
Chapter 1 · From Descriptive to Inferential: What This Course Adds
Probability & Statistics Fundamentals' own Chapter 1 opened with a direction: model known, predict the outcome. Its Chapter 9 flipped that briefly — summarize data you already have. This course goes one step further than either: given only a limited, incomplete sample of data, how confidently can you say anything about the full, unobserved reality behind it? That's inference, and it's the genuinely hard, genuinely useful part of applied statistics.
Descriptive vs. Inferential — A Direction Change
| Probability & Statistics Fundamentals | This course | |
|---|---|---|
| Direction | Known model → predicted outcomes; real data → a summary of that data | A limited sample → a confident, quantified claim about the true, unobserved population |
| Example question | "What's the mean of these 7 response times?" (Ch.9) | "Given these 7 response times, what can we honestly say the true average response time is, across every request that will ever happen?" |
| Core tools | Probability rules, distributions, expected value, descriptive statistics | Sampling distributions, confidence intervals, hypothesis tests, regression, Bayesian updating |
Chapter 9's own worked example computed a mean and median from exactly seven sampled response times. It never asked the harder question this course opens with: how much should anyone actually trust a number computed from just seven data points? That's the entire subject of this course.
Five Concrete Connections to Real Work
| Inferential topic | Where it actually shows up |
|---|---|
| Sampling & sampling distributions (Ch.2) | Every metric ever computed from "the last 10,000 sessions" rather than every session that will ever occur — which is nearly always the real situation |
| Confidence intervals (Ch.3) | Reporting "conversion rate: 5.2% ± 0.4%" instead of a falsely precise single number that hides how much uncertainty the sample size actually leaves |
| Hypothesis testing & A/B testing (Ch.4–6) | Deciding whether a new feature genuinely moved a metric, or whether the observed difference is just ordinary sample-to-sample noise |
| Correlation, causation & regression (Ch.7–8) | Analytics dashboards showing "X correlates with Y," and the classic mistake of assuming that means X causes Y; predicting a continuous outcome from real data |
| Bayesian updating (Ch.9) | Formally revising a belief as new evidence arrives over time — extending Probability & Statistics Fundamentals' own Bayes' Theorem chapter from a single fixed calculation into an ongoing, evidence-accumulating process |
What This Course Builds Directly On
What This Course Won't Cover
A few genuinely related topics stay deliberately out of scope, to keep this course focused on the core inferential toolkit rather than sprawling into adjacent specialties:
- Advanced experimental design — multi-armed bandits, sequential testing, and other more sophisticated A/B testing variants beyond the fixed-sample-size approach Chapter 6 covers
- Machine learning model evaluation — cross-validation, precision/recall, and related ML-specific metrics belong to Data Science & ML's own
ds1/ml1courses, not here - Full computational Bayesian methods — MCMC sampling and similar techniques for genuinely complex Bayesian models stay out of scope; Chapter 9 covers Bayesian reasoning, not Bayesian computation at scale
Where This Course Is Headed
| Chapter | Topic |
|---|---|
| 2 | Sampling & the Sampling Distribution |
| 3 | Confidence Intervals |
| 4 | Hypothesis Testing Fundamentals |
| 5 | The t-test & Comparing Two Groups |
| 6 | A/B Testing in Practice |
| 7 | Correlation vs. Causation |
| 8 | Linear Regression as Statistical Inference |
| 9 | Bayesian Inference & Updating Beliefs |
| 10 | Capstone — Designing and Analyzing a Real Experiment |
Hands-On Exercises
Classify each of the following as belonging more to Probability & Statistics Fundamentals (a known model, or a plain summary of data in hand) or to this course (drawing a conclusion about an unobserved population from limited data): (a) computing the average of 50 measured response times, (b) claiming, from those same 50 measurements, that the true average across all future requests is likely between 190ms and 210ms, (c) given a known fair coin, finding the probability of 3 heads in 5 flips, (d) deciding whether a 2% lift in conversion rate seen in a 2-week experiment reflects a real effect or random noise.
📄 View solutionA colleague says "we don't need confidence intervals — we computed the mean directly from our data, so we already know the real answer." Using this chapter's own descriptive-vs-inferential distinction, explain what's wrong with this reasoning.
📄 View solutionExplain, in your own words, how this course's Chapter 2 (sampling distributions) depends directly on Probability & Statistics Fundamentals' own Chapter 8 (the Central Limit Theorem), and how this course's Chapter 9 (Bayesian updating) depends directly on that same course's Chapter 4 (Bayes' Theorem) — using this chapter's own "not starting from zero" finding.
📄 View solutionChapter 1 Quick Reference
- Descriptive statistics (Fundamentals Ch.9) summarizes data you already have; inferential statistics (this course) uses that data to draw a confident, quantified conclusion about the unobserved whole
- Five direct connections: sampling → any real metric from a subset of data, confidence intervals → honest uncertainty reporting, hypothesis/A-B testing → "is this difference real," correlation/regression → dashboards and prediction, Bayesian updating → revising beliefs over time
- Built directly on two forward references from the sibling course: the CLT (Ch.8 there) feeds sampling distributions here; Bayes' Theorem (Ch.4 there) feeds Bayesian updating here
- Deliberately out of scope: advanced experimental design, ML model evaluation, and full computational Bayesian methods
- Next chapter: Sampling and the sampling distribution