- Home
- Courses
- Data Science
- Data Science Essentials & Machine Learning
Curriculum
- 8 Sections
- 69 Lessons
- 4 Weeks
Expand all sectionsCollapse all sections
- Before You StartIntroduction4
- Module 1: Introduction to Data Science12
- 3.1Principles of Data Science – Data Analytic Thinking
- 3.2Principles of Data Science – The Data Science Process
- 3.3Further Reading
- 3.4Data Science Technologies – Introduction to Data Science Technologies
- 3.5Data Science Technologies – An Overview of Data Science Technologies
- 3.6Data Science Technologies – Azure Machine Learning Learning Studio
- 3.7Data Science Technologies – Using Code in Azure ML
- 3.8Data Science Technologies – Jupyter Notebooks
- 3.9Data Science Technologies – Creating a Machine Learning Model
- 3.10Data Science Technologies – Further Reading
- 3.11Lab Instructions
- 3.12Lab Verification
- Module 2: Probability & Statistics for Data Science21
- 4.1Probability and Random Variables – Overview of Probability and Random Variables
- 4.2Probability and Random Variables – Introduction to Probability
- 4.3Probability and Random Variables – Discrete Random Variables
- 4.4Probability and Random Variables – Discrete Probability Distributions
- 4.5Probability and Random Variables – Binomial Distribution Examples
- 4.6Probability and Random Variables – Poisson Distributions
- 4.7Probability and Random Variables – Continuous Probability Distributions
- 4.8Probability and Random Variables – Cumulative Distribution Functions
- 4.9Probability and Random Variables – Central Limit Theorem
- 4.10Probability & Random Variables – Further Reading
- 4.11Introduction to Statistics – Overview of Statistics
- 4.12Introduction to Statistics – Descriptive Statistics
- 4.13Introduction to Statistics – Summary Statistics
- 4.14Introduction to Statistics – Demo: Viewing Summary Statistics
- 4.15Introduction to Statistics – Z-Scores
- 4.16Introduction to Statistics – Correlation
- 4.17Introduction to Statistics – Demo: Viewing Correlation
- 4.18Introduction to Statistics – Simpson’s Paradox
- 4.19Introduction to Statistics – Further Reading
- 4.20Introduction to Statistics – Lab Instructions
- 4.21Introduction to Statistics – Lab Verification
- Module 3: Simulation & Hypothesis Testing16
- 5.1Simulation – Introduction to Simulation
- 5.2Simulation – Start
- 5.3Lab
- 5.4Simulation – Demo: Performing a Simulation
- 5.5Simulation – Further Reading
- 5.6Hypothesis Testing – Overview
- 5.7Hypothesis Testing – Introduction
- 5.8Hypothesis Testing – Z-Tests, T-Tests, and Other Tests
- 5.9Hypothesis Testing – Test Examples
- 5.10Hypothesis Testing – Type 1 and Type 2 Errors
- 5.11Hypothesis Testing – Confidence Intervals
- 5.12Hypothesis Testing – Demo with R & Python
- 5.13Hypothesis Testing – Misconceptions
- 5.14Hypothesis Testing – Further Reading
- 5.15Hypothesis Testing – Lab Instructions
- 5.16Hypothesis Testing – Lab Verification
- Module 4: Exploring & Visualizing Data4
- Module 5: Data Cleansing & Manipulation4
- Module 6: Introduction to Machine Learning4
- Final Exam & Survey4
Hypothesis Testing – Misconceptions
Misconceptions About Hypothesis Testing
Downloads and transcripts
Video Transcript
- Start of transcript. Skip to the end.
- Okay so now that you think you know something about hypothesis
- testing, it’s time for a quiz.
- So this is a quiz on misconceptions about hypothesis
- testing that’s based on the document that cannot recently
- from the American Statistical Association.
- True or false, p-values indicate how incompatible
- data are with a specified model of the world.
- True or false?
- And of course the answer is true.
- It’s how incompatible your data are with the null hypothesis.
- Next question.
- A p-value measures the probability that the null
- hypothesis is true.
- True or false?
- No way, p-value measures the probability to observe
- something as extreme or
- more extreme than what you did under the null hypothesis.
- I’ll repeat that again, p-value measures the probability that
- the probability to observe something
- as extreme as what you did under the null hypothesis does
- not measure the probability that the null hypothesis is true.
- True or false?
- P-value measures the probability that the data were produced by
- random chance alone.
- And the answer is clearly false.
- I have no idea what it means, random chance alone.
- Right, that’s not it.
- The p-value has got to be thought of with respect
- to the null hypothesis, which is a probability distribution.
- Random chance alone to me means nothing.
- True or false, a p-value below 0.05 is sufficient to base
- scientific conclusions or business decisions or
- policy decisions.
- And in general, the answer to that is false.
- There’s nothing special about that 0.05, right?
- It depends how the study is done and
- whether there’s enough other evidence to make that decision.
- The p-value is a piece of evidence.
- Assuming that somebody didn’t manufacture the data to get
- the p-value they wanted, which people do.
- Obviously, this is totally unethical, but
- people do it anyway.
- So, anyway, the p-value is one piece of evidence.
- True or false?
- A p-value measures how big an effect is.
- A small p-value means a large effect.
- What do you think?
- No way.
- Okay, so here’s an example, n=10 million.
- I have 10 million students.
- Say they all go to a tutoring session.
- The tutoring session improved their score
- over the population mean by only say five
- hundredths of a point, five hundredths.
- Let’s say that the population mean is 80 and
- after they go to the tutoring session, the the score
- of the students becomes only slightly, slightly higher.
- And I could calculate the p-value for testing whether
- the tutoring session improves their score above 0,
- and I might get that the p-value is 0.0001.
- So highly significant.
- I’m really sure that the tutoring sessions improved
- their score.
- But whoop dee do, yeah, it improved their score, and
- I’m really sure of it.
- But it only improved the score by a tiny little itty
- bitty drop.
- Okay, so don’t get fooled by this one.
- The p-value is not the size of the effect, no way, no how,
- false.
- True or false, a p-value tells you how important a result is.
- And the answer to this one should be obvious.
- If it doesn’t tell you how big the effect is,
- then how can it tell you how important the result is?
- The p-value is the first line of defense against being fooled by
- random chance.
- They’re very helpful and
- I suggest you use them with caution.
- End of transcript. Skip to the start.