This is what true education of the future looks like. Simple, creative, fun to watch, highly educational, and very graphic. And a great job by Adriene Hill and the entire team for the quality of the content. Keep it up.
Grace Sophia stats gets super cool! Eventually u'll see the connection between all sorts of stats that initially seemed to be completely different, like ANOVA and Linear Regression.
Jacobo OH, it's definitely my hardest AP. But it's also my most fun and most rewarding. I'm only taking four (and one was a semester class I don't have this semester) so I'm just going to say "I'm sorry". Six AP classes is a hell of a lot.
10:06 : I think it should be mentioned here for the more mathematical-minded folks that that's a difference between what's called 'statistical probability' and 'theoretical probability', and a general principle is that statistical probability approaches the theoretical probability as the number of trials approaches infinity (which could be thought of as a limit at infinity).
Is there any way that you could release transcripts, outlines, and/or notes for your videos? Especially the math-heavy ones. I think it would be easier to learn the information if there were a guide
i had a course on statistics about two years ago, and wow, i'm really thankful for this course being here, because it's nice to be reminded of these things
I really like the animations in this video, especially that you can then see how changes in the data distribution affects the plot. This way it starts to make more sense to me.
When you were explaining how standard deviation changes the shape of a normal distribution, shouldn't the numbers on the horizontal axis stay where they are instead of stretching and squishing with the standard deviation? Because I think that otherwise the shape of the distribution doesn't really change. Please, let me know if I'm missing something.
Fernando Franco Félix I suggest sleeping on it. When my code doesn't work I usually fix it easy in the morning. Once I was so tired I didn't realize I was trying to run correlations on zero vectors. XD
Fernando Franco Félix - Maybe somebody got lazy and left out a higher order term in estimating a cross section that you’re using... just kidding; nobody would do something like that !
5:13 if lots of people died young during the middle ages, wouldn't that move the curve/average for "age of death" further left, thus moving the skew (tail) to the right, i.e., right skewed? Right now with the average age of death around 80, it's left skewed.
The timing of this title was perfect because I totally read it as "The Shape of Water" when I glanced at it really quickly. This comment will be irrelevant by the end of the week... because no one will get it.
I really enjoy these videos! My only comment/complain/concern is that a lot of your examples and expression are North American. I am from South Africa and was fortunate enough to spend a year in the USA but other English speaking people from around the world will not be able to understand expressions like lollygagging etc. Otherwise 5 star!
Am I the only one for whom the 'knobs and dials on the machine' metaphor doesn't work? I like this series so far (and love Adriene! Crash Course Economics is a real good series y'all), but this video was a little confusing for me.
I rolled a single 20-sided die 10750 times. The 1 came up 2.7% below the mean and the 20 1% below. The 9 and 15 each came up 0.9% above the mean. Comparing the actual numbers of 15 and 1 that's 639 to 247. Out of 10750 that seems like a big difference to me. Is this enough samples to say that the die is an unfair die? Throughout the process I watched the distribution. It stayed pretty consistent once I got to around 1000 rolls. Counting the 1 as the bottom and 20 as the top, the 9 and 15 are on the middle row directly above the halfway point and are almost opposite each other. The full data 1 247 -2.7% 2 510 -0.3% 3 603 0.6% 4 503 -0.3% 5 587 0.5% 6 559 0.2% 7 547 0.1% 8 513 -0.2% 9 636 0.9% 10 512 -0.2% 11 551 0.1% 12 577 0.4% 13 578 0.4% 14 500 -0.3% 15 639 0.9% 16 482 -0.5% 17 627 0.8% 18 548 0.1% 19 599 0.6% 20 432 -1.0%
Firstly, I'm curious: when and why did you roll a die 10750 times? How long did that take you? Was it a physical die or a digital random number generator? Secondly, I couldn't resist running some chi-square tests. An overall test of your whole dataset (where the expected frequency of each number is 10750/20=538) generated a whopping chi-square value of 271. That's a measure of how far the observed distribution is from the expected uniform distribution, and with 19 degrees of freedom, we get a p-value of, uh... about one to the minus 46. This suggests that it would be *extremely* unlikely for a fair die to give such a wonky distribution of values. I was curious to what degree this result was affected by the very low number of 1s, so I tested a couple of number frequencies individually against the combined frequency of the other numbers. Turns out, even getting this many nines (observed 636:10114; expected 538:10212) seems very unlikely (chisq=19; df=1; p
I was going to roll 10,000 but I was bored and rolled some more. I rolled it off and on for several days while I watched UA-cam and Netflix. I wanted to see how fair it was because I bought some new dice, had watched some Matt Parker videos about dice, and thought it would be fun to try out the 20-sided one. (Yep, I don't have much to do; I'm disabled and sit around a lot) Thanks for doing the statistics. It's been over 20 years since I took statistics in college, and my memory is horrible; I didn't know where to start.
That bimodal distribution with the Boston Marathon had two peaks, one for men one for women. It was even labeled on the graph. Oops. Props for the gamers' dice though!
@4min17sec, is 'standard deviation the average distance between any point and the mean'? I thought MAD (mean absolute deviation) is the definition of that?
5:14 "age at death during the middle ages is left-skewed... cause lots of people died young..." Is "left-skewed" a slip of the tongue? Since more people died young, the median should be to the left of mean, assuming ages lie on the x-axis from 0 to 100, which gives the right-skewed distribution.
Is there going to be a crash course maths? I feel it's weirdly missing, given the large amount of science and other things that rely on a good basic understanding of it and the connected subjects such as discrete maths and algorithmic maths Not sure if this is the best place to ask this though...
Is there a test we can make on the data to know if a bimodal distribution is actually an overlap of two distributions, or do we need to analyse each case separately?
This is just a guess, but since Q refers to the proportion of population elements that do not have a particular attribute (in this case, Awesomness?) Then DFTBAQ reads Don't Forget To Be Awesome, All of you people who are currently NOT AWESOME. :) Presumably the Already Awesome people didn't forget, and thus don't need the reminder.
Somebody reply plz ...kinda off topic but...since we don't count every opinion to have an idea of a the actual picture , and we just take into account the majority...That means the majority is what must be considered? For example 1% < 99%
...2 min. vs 4 min. (bimodality) sounds more like frequency-doubling-a nonlinear process with harmonics... there's often more important, extractable, shaping information, analysts ignore as massaging-too-much after tainting by predecisory mode-information-politics...
What does she mean it 'generates random numbers'??? and what does she mean it 'generates the number of leaves on a tree?' does she mean the machine counts the leaves or theoretically generates the leaves? I'm in post grad and this makes no sense
It just means that it's continuous. People can be 176cm, 176.1cm, 176.11cm, 176.2cm, 176.223cm, 180cm, 180.5cm, 180.6cm etc. There's just so many values height can take that it's infinite.
A “theoretical” roulette wheel yields a uniform distribution over final slots but check out the book The Eudaemonic Pie for an entertaining story about a group who took advantage of the slight imperfections associated with any real roulette wheel.
watching informative things without meaning to is how you acquire knowledge that one day you'll say "i have no idea how i know that" about idk if that sentence made sense.
A company that discriminates on the basis of age, race, gender, or similar will possibly not be as competitive as one that doesn't - companies need the best people, regardless of age, race, gender, or similar. So if a company _seems_ to be hiring one group more than another, one must look at their reasons; and not simply assume they're discriminating.
Causality is complicated! But, data can be used to explore the question. However, I don't think asking their reasons will give the correct answer either. Companies that discriminate seem to be the type to invent reasonable sounding excuses to obfuscate.
I suspect that in this (litiginous) environment, all such reasons will be included in the H.R. records, just in case someone questions the results. And all statistical examinations of such decisions should probably include the findings of other research into physical strength and endurance, I.Q., or whatever else might influence the hiring managers' decisions. Who would question the manager of an NBA team for hiring way mostly dark-skinned people? Or who would question the manager of an IT team for having a much higher percentage of Chinese than the percentage of Chinese in the local community?
Aguila: That's the opposite of what I've just said. I'm well aware of the inequality between the different "races, ages, and genders". It's known that the Chinese are generally smarter than most other nationalities (excluding the sub-groups like Ashkenazi Jews who are generally the smartest), and that the African groups generally fill the other end of the spectrum, but are the best at some sports. And it's precisely because of this that every person must be treated on their own merits and for the current need. It's ridiculous to claim that "all men are created equal" when the opposite is apparent to anyone who has observed his fellow man. Of course, that claim in the US founding documents was probably about being equal in the pursuit of certain goals, and not about everyone being able to press 150 kg.
IQ, of course - the most valid measure there is - see the talks by people like Jordan Peterson for details. The only people who don't accept it seem to be those who score low. But it correlates much better than any other measure with economic success in any intellectual pursuit. The mean IQ of B.S. graduates in Physics, (the hardest of the sciences) for instance, is over 130. Engineering and computer science are just a bit less, and the lowest are degrees not much more challenging than basket-weaving.
![The 6 MUST-KNOW Statistical Distributions MADE EASY [4/13]](http://i.ytimg.com/vi/V3Qxj2C7rP0/mqdefault.jpg)
This is what true education of the future looks like. Simple, creative, fun to watch, highly educational, and very graphic. And a great job by Adriene Hill and the entire team for the quality of the content. Keep it up.
This fundamental stuff is so important, yet overlooked in most academic statistical practices
I love how this series mixes the answers to big picture "why?" with the more pixelated "what?" and "how?" questions.
I'm taking Stats this year, and we're on significance tests, but I'm watching all of these too because I think Stats is really, really cool.
Grace Sophia stats gets super cool! Eventually u'll see the connection between all sorts of stats that initially seemed to be completely different, like ANOVA and Linear Regression.
Just took a test on that today lol
what human are you? I toke ap stats as a joke and I got in, Im taking six aps this year is stats is one of the hardest
Jacobo OH, it's definitely my hardest AP. But it's also my most fun and most rewarding. I'm only taking four (and one was a semester class I don't have this semester) so I'm just going to say "I'm sorry". Six AP classes is a hell of a lot.
10:06 : I think it should be mentioned here for the more mathematical-minded folks that that's a difference between what's called 'statistical probability' and 'theoretical probability', and a general principle is that statistical probability approaches the theoretical probability as the number of trials approaches infinity (which could be thought of as a limit at infinity).
Is there any way that you could release transcripts, outlines, and/or notes for your videos? Especially the math-heavy ones. I think it would be easier to learn the information if there were a guide
If you click the 3 dots to the right of the thumbs up/down and share buttons, there is an option to open a transcript.
oh my god.... thank you.. i wish i knew this earlier but thank you!! :D
This course is one semester late and could have helped save me from getting a D+ last semester in my Stats class. But D's get degrees, baybee!
Since when do D's get degrees? Anything below a C average was below a 2.0 GPA, and put you on academic probation at my university.
i had a course on statistics about two years ago, and wow, i'm really thankful for this course being here, because it's nice to be reminded of these things
Thanks for taking the time to make the video. Sadly, it lost me at the Standard Deviation.
I really like the animations in this video, especially that you can then see how changes in the data distribution affects the plot. This way it starts to make more sense to me.
"samples, shapes and their shadows" .. superb writing skills.
Crash course: music theory
Elijah Cook Yes!
i'd like a course on music theory that uses mathematical language instead of the horrible nomenclature musicians use. and chromatic notation.
sofias. orange I couldn't agree more
YES PLEASE
I agree. As a JH/HS instrumental teacher I love teaching applied music theory but I don't always give it the time it deserves.
When you were explaining how standard deviation changes the shape of a normal distribution, shouldn't the numbers on the horizontal axis stay where they are instead of stretching and squishing with the standard deviation? Because I think that otherwise the shape of the distribution doesn't really change. Please, let me know if I'm missing something.
Great video - really helpful. Like the touch with the bell curve /box plots/pie charts in the background!
I'm a particle physicist and my simulation ins't working and I'm tired so I'm gonna watch this video to see if they mention something I'm overlooking
Fernando Franco Félix I suggest sleeping on it. When my code doesn't work I usually fix it easy in the morning. Once I was so tired I didn't realize I was trying to run correlations on zero vectors. XD
Fernando Franco Félix - Maybe somebody got lazy and left out a higher order term in estimating a cross section that you’re using... just kidding; nobody would do something like that !
well just pretend to be a particle. Then you know how.
Play this at low volume while you sleep:
Particle Man -- They Might Be Giants
Have you rebooted the flux compensator? Happens to me every time!
my fav part of statistical mathematics is how often she says knobs.
Too bad this video didn't feature Ed Sheeran creepily stalking the entirety human population because he likes the shape of their data.
Probably he'd stalk a decent random sample of the population since the sample should have a similar distribution.
World History With Dan 😂
Wouldn't that be Grint doing the stalking?
5:13 if lots of people died young during the middle ages, wouldn't that move the curve/average for "age of death" further left, thus moving the skew (tail) to the right, i.e., right skewed? Right now with the average age of death around 80, it's left skewed.
The people who are old enough to have watched Animaniacs are going to be the ones doing better on the countries of the world test.
Alexander Roderick What about Freakazoid, what can the people who've watched it do?
What about the people who didn’t have a childhood what can they do?
Suvi-Tuuli Allan they can say Candle Jack and then suddenly dissa
Animaniacs has been remade
Ahahahah
Today I learned that male marathon runners = professionals and female marathon runners = amateurs
Love the title. I see what you did there.
What does DFTBAQ mean?
Dont Forget to be Asking Questions
Thank you! Already put into practice and shown to be beneficial
Thanks
She is so pretty.
am i the only one taking a statistics test tommorow???
Mary Joanne Miranda No sis, me too🙃
no
@@nikolakis2001 same
I have a quiz tomorrow
Same I have an AP screening test
The timing of this title was perfect because I totally read it as "The Shape of Water" when I glanced at it really quickly.
This comment will be irrelevant by the end of the week... because no one will get it.
I don't get it.
You're videos are amazing and so easy to understand. Thank you! (As I work through each one by one....)
I really enjoy these videos! My only comment/complain/concern is that a lot of your examples and expression are North American. I am from South Africa and was fortunate enough to spend a year in the USA but other English speaking people from around the world will not be able to understand expressions like lollygagging etc. Otherwise 5 star!
Am I the only one for whom the 'knobs and dials on the machine' metaphor doesn't work? I like this series so far (and love Adriene! Crash Course Economics is a real good series y'all), but this video was a little confusing for me.
parameters of the normal distribution function
from min6 is good
Please cover linear regression!!
This is a decent resource for regression analysis!
If the variance is low, that means that data is squished closer to the mean, so why is pointiness bad (leptokurtic)?
I rolled a single 20-sided die 10750 times. The 1 came up 2.7% below the mean and the 20 1% below.
The 9 and 15 each came up 0.9% above the mean.
Comparing the actual numbers of 15 and 1 that's 639 to 247. Out of 10750 that seems like a big difference to me.
Is this enough samples to say that the die is an unfair die? Throughout the process I watched the distribution. It stayed pretty consistent once I got to around 1000 rolls.
Counting the 1 as the bottom and 20 as the top, the 9 and 15 are on the middle row directly above the halfway point and are almost opposite each other.
The full data
1 247 -2.7%
2 510 -0.3%
3 603 0.6%
4 503 -0.3%
5 587 0.5%
6 559 0.2%
7 547 0.1%
8 513 -0.2%
9 636 0.9%
10 512 -0.2%
11 551 0.1%
12 577 0.4%
13 578 0.4%
14 500 -0.3%
15 639 0.9%
16 482 -0.5%
17 627 0.8%
18 548 0.1%
19 599 0.6%
20 432 -1.0%
Firstly, I'm curious: when and why did you roll a die 10750 times? How long did that take you? Was it a physical die or a digital random number generator?
Secondly, I couldn't resist running some chi-square tests. An overall test of your whole dataset (where the expected frequency of each number is 10750/20=538) generated a whopping chi-square value of 271. That's a measure of how far the observed distribution is from the expected uniform distribution, and with 19 degrees of freedom, we get a p-value of, uh... about one to the minus 46. This suggests that it would be *extremely* unlikely for a fair die to give such a wonky distribution of values.
I was curious to what degree this result was affected by the very low number of 1s, so I tested a couple of number frequencies individually against the combined frequency of the other numbers. Turns out, even getting this many nines (observed 636:10114; expected 538:10212) seems very unlikely (chisq=19; df=1; p
I was going to roll 10,000 but I was bored and rolled some more. I rolled it off and on for several days while I watched UA-cam and Netflix.
I wanted to see how fair it was because I bought some new dice, had watched some Matt Parker videos about dice, and thought it would be fun to try out the 20-sided one. (Yep, I don't have much to do; I'm disabled and sit around a lot)
Thanks for doing the statistics. It's been over 20 years since I took statistics in college, and my memory is horrible; I didn't know where to start.
no
That bimodal distribution with the Boston Marathon had two peaks, one for men one for women. It was even labeled on the graph. Oops. Props for the gamers' dice though!
That was not the point
good presentation for what was covered, but was really looking for a lot more distributions like chi squared, poisson etc
I decided I must come here because after 6 hours I have completed a single statistics question.
Wouldn't age of death in the middle ages be right skewed? Wouldn't all of the older ages be on the right, and therefor the tail be on the right..?
@4min17sec, is 'standard deviation the average distance between any point and the mean'? I thought MAD (mean absolute deviation) is the definition of that?
I could totally be mistaken but did she make a mistake when she said the mortality age was left skewed. Should it not be right skewed?
yh ur right, it should be right skewed instead.
Wish they made these vids earlier so it could’ve actually helped me with AP statistics back in fall smh
Great video! I'm eager for seeing the other distributions like poisson and bernoulli.
I love looking at the shape of data!
No one is gonna be getting 100% on that name all countries test, hell I've been trying to get there for 2 years and I can only hit about 180/195
please teach us art of electronics for begginers
5:14 "age at death during the middle ages is left-skewed... cause lots of people died young..." Is "left-skewed" a slip of the tongue? Since more people died young, the median should be to the left of mean, assuming ages lie on the x-axis from 0 to 100, which gives the right-skewed distribution.
Ive been watching this video until the end.. And.. I suddenlly remember.. That i need to take my clothes to dobi
Hey Y'all
Please add course of agriculture..
CrashCourse mathematics required.
I just took a data analysis test today.
whered you take it, to the ohhh watenmdfaikkv c
Proud to say I could name most Pokemon and countries in the world
I just realized that the shelf behind her is shaped like a bell curve... XD
I want that tiny pillow.
can someone please tell me where that normal curve pillow on the table is from?!
Decent overview, but perhaps too simple.
Thanks!
All I learned from this.
Roulette has 38 slots.
I expected to hear the concept of kurtosis
Is there going to be a crash course maths? I feel it's weirdly missing, given the large amount of science and other things that rely on a good basic understanding of it and the connected subjects such as discrete maths and algorithmic maths
Not sure if this is the best place to ask this though...
William Walls crash course is produced by pbs. Pbs has another channel called infinite series that deals with math.
Is there a test we can make on the data to know if a bimodal distribution is actually an overlap of two distributions, or do we need to analyse each case separately?
Thank you so much!!!!! This really helps
Uniform distribution often refers to continuous uniform distribution. Dice makes Discrete Uniform Distribution
thanks
That’s harder then the physics one.
The marathon statistics say, male and female, not professional and amateur.
Good for refreshing your knowledge. But I can imagine this is terrible video for learning the first time.
Crash Course Music! The science of harmonics and it's application in all the instruments as well as it's evolution from ancient times up to today!
I just realized the shelves on the wall are box-whiskey plots :P
She reminds me of rudy macunso
Q?
I'm Wondering that too.
ThePstjtt Don't Forget To Be Asking Questions.
It could have something to do with Lgbtq but I have no clue
This is just a guess, but since Q refers to the proportion of population elements that do not have a particular attribute (in this case, Awesomness?) Then DFTBAQ reads Don't Forget To Be Awesome, All of you people who are currently NOT AWESOME. :) Presumably the Already Awesome people didn't forget, and thus don't need the reminder.
Ken Kopin that would be way cooler than just Don’t forget to be asking questions which is what she said in the last video
No lolly-gagging
I like the bamboo histogram on the table
Somebody reply plz ...kinda off topic but...since we don't count every opinion to have an idea of a the actual picture , and we just take into account the majority...That means the majority is what must be considered?
For example 1% < 99%
...2 min. vs 4 min. (bimodality) sounds more like frequency-doubling-a nonlinear process with harmonics... there's often more important, extractable, shaping information, analysts ignore as massaging-too-much after tainting by predecisory mode-information-politics...
Would you like to suggest any material or reference to follow along with your videos?
I wonder what’s the data on school loans and high earning jobs. Do high earners pay off their student debt(early) or do they even have payments?
Are there any specific exercises to solve or some related question sets for each episode or topics?
What does she mean it 'generates random numbers'??? and what does she mean it 'generates the number of leaves on a tree?' does she mean the machine counts the leaves or theoretically generates the leaves? I'm in post grad and this makes no sense
Crash course anthropology?
Just ask the bank
where is the Arabic sub. for the rest of the corse ?! ):
i understand that correlation does not mean causation, but does that also mean that if there is causation then correlation is implied?
I think this lady in the video looks like Lexi Walker.
The normal distribution is also called the gaussian distribution
DFTBAQ
Someone please explain to me how height is infinite??
It just means that it's continuous. People can be 176cm, 176.1cm, 176.11cm, 176.2cm, 176.223cm, 180cm, 180.5cm, 180.6cm etc. There's just so many values height can take that it's infinite.
A “theoretical” roulette wheel yields a uniform distribution over final slots but check out the book The Eudaemonic Pie for an entertaining story about a group who took advantage of the slight imperfections associated with any real roulette wheel.
I totally did not understand the machine example.
I'll have to watch again.
The marathon time distributions were for amateur and pro? Because the label said male and female.
If a distribution is normal does that mean it has only one mode? Can it be bi-modal and normal?
WHOOOHOOO FIRST VIDEO I AM EARLY ON OMG YAY I'm THE 7TH COMMENTER AND 29th LIKER GUYSSS I DID IT
she looks like a young angela merkel
Aaaw, a plush normal distribution!
DFTBA... Q (lol)!
Why am I waching this. lol
because it is well made.
watching informative things without meaning to is how you acquire knowledge that one day you'll say "i have no idea how i know that" about
idk if that sentence made sense.
in the example of the two different types of marathon runners -- why are amateurs labeled as "women" and those who compete as "men"??
do you see that second yellow peak? think before you comment
I'm finally early WOOOOOOHOOOOOO
Notice the bamboo shoots like a normal distribution
ترجمه عربى لو سمحتوا
A company that discriminates on the basis of age, race, gender, or similar will possibly not be as competitive as one that doesn't - companies need the best people, regardless of age, race, gender, or similar. So if a company _seems_ to be hiring one group more than another, one must look at their reasons; and not simply assume they're discriminating.
Yes it wouldn't prove discriminatory practices but it would be a first step towards a causal argument.
Causality is complicated! But, data can be used to explore the question. However, I don't think asking their reasons will give the correct answer either. Companies that discriminate seem to be the type to invent reasonable sounding excuses to obfuscate.
I suspect that in this (litiginous) environment, all such reasons will be included in the H.R. records, just in case someone questions the results. And all statistical examinations of such decisions should probably include the findings of other research into physical strength and endurance, I.Q., or whatever else might influence the hiring managers' decisions. Who would question the manager of an NBA team for hiring way mostly dark-skinned people? Or who would question the manager of an IT team for having a much higher percentage of Chinese than the percentage of Chinese in the local community?
Aguila: That's the opposite of what I've just said. I'm well aware of the inequality between the different "races, ages, and genders". It's known that the Chinese are generally smarter than most other nationalities (excluding the sub-groups like Ashkenazi Jews who are generally the smartest), and that the African groups generally fill the other end of the spectrum, but are the best at some sports. And it's precisely because of this that every person must be treated on their own merits and for the current need. It's ridiculous to claim that "all men are created equal" when the opposite is apparent to anyone who has observed his fellow man. Of course, that claim in the US founding documents was probably about being equal in the pursuit of certain goals, and not about everyone being able to press 150 kg.
IQ, of course - the most valid measure there is - see the talks by people like Jordan Peterson for details. The only people who don't accept it seem to be those who score low. But it correlates much better than any other measure with economic success in any intellectual pursuit. The mean IQ of B.S. graduates in Physics, (the hardest of the sciences) for instance, is over 130. Engineering and computer science are just a bit less, and the lowest are degrees not much more challenging than basket-weaving.