Thank you so much for this video. I've been passionated for stats since my first class on it, but I often question myself if I should keep pursuing this major giving my dificulteis with math. I’m definitely not the quickest with numbers but I do get the concepts. I also have a hard time memorizing facts but on the other hand I have a strong logic, so I am usually able to draw conclusions if I have time for it. It’s challenging because I see most of my classmates looking at problems and solving them within seconds and I take so long. Anyway, your explanation between the subjects is very encouraging. I’ve always loved philosophy, psychology, anthropology, and I feel like stats is the tool to put all these subjects together to better understand the world and our society so, thanks again for the encouragement.
There are a lot of statisticians out there that are lousy at math. If you know what needs to be done you can ask the computer to do the calculations. There is a branch of stastistics called "theoretical research" that is VERY math intensive, and you've gotta be good at math to succeed there, but there's all sorts of applied research and consulting options that don't require you to know all the math tricks. If you love science and get math mixed up there is a large area of statistics that is perfect for you.
@@sdcstats There are statisticians and then there are mathematical statisticians. The latter are more competent. There are two kinds of statistics: 1. Descriptive 2. Mathematical Anyone can learn descriptive statistics. Not anyone can learn mathematical statistics.
I am really glad I came here to understand the difference between statistics and math. The internet is filled with answers that are like the first example you used and it's nice to know it's not so simple. I am a freshman undergraduate looking at both a math and statistics major. I really like how applicable statistics is to the real world. Not that you said mathematics isn't. But maybe just not in the ways that I am looking for.
Thank you Dr. Crawford for your wonderful explanation. I especially enjoyed your breakdown of an intro stats class into math and statistical concepts. Could you please recommend an intro textbook that adequately addresses the topics you listed under statistics? Keep making videos! You are saving the world!
Well my favorite is probably "Introduction to the practice of statistics" by Moore and McCabe. The book by Agresti and Franklin "Statistics the art and science of learning from data" isn't bad either (a little more technical, less on the pictures). A lot of other teachers swear by the authors Utts (who keeps things quite simple) or Devore (who is a bit more advanced). If you want a good intro book that is advanced enough for most secondary stat classes then "The Statistical Sleuth" by Ramsey/Schafer is a good pick.
So good. Students should watch it in the beginning, being kind of confused, and then at the end of semester so they can click with a concept more. It's really good put together.
Excellent video! The fact that students who approach stats as if it were a math class do poorly is testimony to the way math is too often taught, which is sadly in a step-by-step algorithmic computational fashion, or at least that's how many students try to get through it. Instead, they should see math as rigorized thinking and precise expression of relevant philosophical concepts such as form, pattern, distance, closeness, continuity, etc. The "step by step procedures" in textbooks is partly responsible for this. So they learn to manipulate formulas, etc., using this approach, pass the course, then get to stats and think they can do the same to succeed. But that kind of thinking doesn't work in stats (assuming the course is taught well), because of all the reasons Scott mentions. Stats, when combined with research and science, requires many inputs to reasoning through a problem, and in scientific applications, it can prove very difficult figuring out optimal designs and appropriate analyses. Not only can it be difficult verifying assumptions, etc., but even determining whether your variables are best measured on a pseudo continuous scale or categorical can be very difficult. And, even once you've conducted your analyses and answered your hypotheses, determining what you can vs. cannot conclude scientifically can be very challenging. Most students greatly overestimate what can be concluded from a statistical model. What is usually more relevant to drawing strong scientific conclusions is the underlying research design. As Scott discusses, it's never as simple as simply running a statistical test. When you're working with statistics, you're usually working with science at the same time, and all the things Scott mentions has to be considered in your decision-making and in drawing conclusions from your analytical method. You have to be well-versed in the field in which you are applying the statistics so you know the types of conclusions that can vs. cannot be drawn on a substantive level.
Fantastic explanation. Superb editing. Transition shots taken so meticulously that the edit does not seem jump cut (except few places). The video making helps the topic to be more interesting. Thanks for the video!
I love how you explain everything... I have a question. I am planning to get my masters degree in statistics science... however I suck at math. but im good in marketing and understanding what people are looking for and why through patterns in habits... do you think this is a right path for me? please explain
There are a lot of very successful good statisticians out there who aren't very good at math. Thanks to technology you absolutely can become a great data scientist without great math skills. That being said - there is at least a certain amount of math that is expected: for example if you understand how an integral works then you'll be more comfortable with what the computer is doing when you get probabilities under some distribution. You've got to be comfortble with fractions, and how to read a polynomial equation. But you don't need much beyond that. As for getting a degree without great math skills - now that depends on the program. You'd want to ask questions like "Is this program heavily theoretical or more applied?" The more theoretical it is the more expectations they will have of good math skills. Still no matter where you go there's going to be a theory class, and you'd have to struggle through it. The good news is that after fighting through it if you feel like your theory skills are still weak you can still be a great statistician.
Brilliant, just the stuff we were looking for - thank you for your service! ... Unfortunately, a lot of the elderly people +50 have no idea what they are doing, while living ... so we have to teach the following generation. And a lot of the elderly people in the deep state seem to be just cowards, they collect the data on the general public - and, with some help of some younger guys, they know what they are doing, which is exactly what you showed to us. Just applied on the general public, all-time, everywhere. And this is why they are cowards with fear to loose control: A public being as smart and as well informed as they are would help to develop the societys forward. There is no use for secrets - cause our society's have to use their brains to the max. Maybe together with a good cup of tea from time to time, yes.
Great video! Great orator skills. Great video editing skills. Great teaching...order of information, explaining a new concept by referencing/leveraging concepts that the student already knows and building off of it. GRRRR! Why can't all teachers be this good. Can we just let STEM individuals restructure everything in society! Get rid of incompetent narcissists who can't innovate or solve a problem to save their life so they instead paywall things like natural resources and education and then go on to artificially create self-worth by creating and perpetuate myths like "engineers lack social skills."
I'm doing a mathematics degree in UK and theres a unit for mathematical statistics and it's VERY mathsy and doesn't deal with much raw data an anlaysis. But I guess this is more the theory behind it so maybe thats why
We are using different sciences to solve world problems. Statistics is a science to add improvements where there is no permanent solution. Math is a science which can resolve problems permanently and their should not be any room for improvements.
What does an arithmetic mean tell one? [A] Nothing [B] It's just an average [C] It's easy to compute: just add up the numbers and divide by the total [D] The arithmetic mean is only valid when data can be redistributed within a given set, meaning that in such cases, there are never any outliers. [E] Class arithmetic means are useless because students cannot share their grades Which is the best description of what is an arithmetic mean?
I wanted to comment on your other response, but I don't see it anymore. I've had people say their comments got lost, and I don't know why. You made a good point that knowing mathematics is going to be a key ingredient to high level statistics. I made this video because the place where I used to work was combining the mathematics and statistics departments, and many of the math faculty couldn't understand how we were different. They really believed we just taught "applied math". It was a big deal when it came time to evaluate my job performance because they were so different. When I showed them how many research projects I had been associated with they accused me of "moonlighting". I made this video in part to help them understand how our statistics classes were very different than "applied algebra". It was a wild time, and I'm glad I moved back into a separated statistics department. Your question above is a fun one - each answer could be valid depending on what you're trying to do. There certainly are examples where the arithmatic mean isn't useful, but others where it very much would be. Class means always drive me nuts - students will take a test and ask "what was the average" and I cringe thinking "It was part of the exam to ask why you wouldn't want the average in this case...." In engineering the mean is called the "first moment" because you're collapsing the entire data set into one value. In physics they call it the center of gravity. Knowing the center of gravity is just as useful as knowing the mean. There's scenarios where it would not be sufficient, but depending on the question you're asking it might be a great way to simplify the calculations. If I really had to pick an answer I think I'd go with C.
@@sdcstats The correct answer would be [D] because that's exactly what happens when you do that which is described in [C], ie, you redistribute the values of each datum so that they are all equal. An arithmetic mean is useful when it makes sense to redistribute the values so they are all equal. On the other hand, it makes no sense to compute an arithmetic mean when it makes no sense to redistribute the values. For example, an arithmetic mean of peoples' height is useless, because you can't chop them all up so that they all have the same height. This is the same as taking a class grade - it tells you nothing about a student's performance. In a class of 10 students, the following scores are recorded: 30,30,30,30,30,30,30,90,100,100 The mean is 50 implying that most students would have passed, but this is obviously FALSE. 7 students failed. Now for a useful mean: Suppose there are 3 cities that each require 2 million cubic meters of water each month. If we are at the end of a month, how can we tell if the current water reservoirs have sufficient water for the next month? That's right! We determine the arithmetic mean of the three cities reservoirs that are connect by conduits. If the mean is 2 or greater then we are done. Otherwise, we will have a water shortage. City A has 1 million cubic meters City B has 2 million cubic meters City C has 3 million cubic meters In the above scenario, the mean tells us that we have enough water if it is redistributed or shared between the cities. In fact the mean value theorem which is the reason calculus works, is about an arithmetic mean. Make sense?
@@sdcstats What I am telling you is that you can know beforehand if it is logical or not to calculate an arithmetic mean. You can know. If redistribution of values makes sense, then the arithmetic mean will be valid with no outliers or anything else unexpected. If redistribution of values does not makes sense, then no need to calculate an arithmetic mean.
You don't need to be good at maths to do well in statistics. You can leverage your math skill with computer science skill (basically asking the computer to do the math for you) or a business analyst direction (knowing how to be part of a team where some do the math and you do the planning). Not wanting the heavy math does limit some options (like a theoretical PhD) and not every school can offer a direction that has less math, but those options certainly do exist and can lead to happy careers.
I might major in Statistics. It sucks that I need a Master's degree to be considered competent. Like what? Four years isn't enough to cram all those math classes?
There are some schools (like A&M) where a Bachelors in Statistics means the student is suprisingly competent. But many schools treat the bachelors in statistics as a math degree with some extra probability and basic statistics vocabulary. Because of that businesses often view a bachelors in statistics as "They know how to handle data, and they're good with complex instructions, so they can be number crunchers." On the other hand masters students are seen as "They understand statistical methods and have some experience with real application of the common types of analysis", so they do the analysis after the bacehlor has cleaned the data. Then you have PhD expectations: "They either know all the methods, or can figure them out with a little bit of time. They could go beyond what's been done and do something new or unique if we needed.", so they meet with the CEO to talk about what is needed, plan how to get the data, and tell the master's level hire what analysis to do after the bachelor has cleaned the data.
1) No one is "bad" at maths. It just means it will take you longer to learn than others, but it takes everyone time, and if you're slower you CAN still learn it. 2) Find good friends. Please who do see mathematical things quicker can help you get throught the tougher things. Don't be a leech, no one wants someone who pretends to be a friend just to get answers. But be kind and friendly to others, work hard, and they will recognize you are a good person and help teach you. 3) Aim towards classes that are more applied and less theoretical. That depends on what school you go to, but often classes like regression, categorical, nonparametric, and data mining will emphasize the applied side. Often classes like multivariate, time series, spatial, bayes (advanced) will be more theoretical. That school dependent, though, so ask around. You'll also find certain teachers tend to emphasize the theory/application more, so gravitate towards the teachers that follow your style. 4) Be patient. There is no shame in repeating a class, especially if it's a theory heavy course. You can make it, and you can be a good statistician without good maths skills. If it takes you longer to graduate no one will care 3 years later. Always have the attitude of "Wow, this course is hard, but I can do it!"
It will depend on your skills and your career goals. If you're a very systematic formulaic thinker you'll probably like math better. If you're the type to see the big picture and want to do concrete applied things then you'll like statistics better. If you're planning on graduate schoool those programs (pretty much all of them) like math majors, because they can focus on teaching their subject instead of the math tricks they use. Statistics will teach you their subject while talking about all the others at the same time (engineering examples, sociology examples, biology examples, etc.) If you want to be in academia (a teacher or a professor) then you can pick whichever subject interests you most. If not, then math should be a jumping off place where you eventually move into the subject you're interested in (towards your career). Statistics by itself leads into lots of great careers that are in demand now and expected to stay in demand for a long time. You wouldn't have to switch majors/careers later, and you can steer towards the type of statistics that deals with the field of science you're most interested in.
Well my opinion will ovbiously be biased towards statistics, but it does sort of depend on where your goals are. If you want to be a theoretical researcher or go for a PhD at a top tier research institution go for math. If you're aiming for a business, medical, financial, or commercial occupation then you'll want statistics. If you're aiming for cyber security or database management or big data machine learning algorithms then you might prefer computer science. And all of these jobs have shades of grey and overlap. What many undergraduate students do is pick the major (math, statistics, computer science, business, or econ) that they enjoy and are good at, and then after 4 years of focused study they have a better idea of what career path they are interested in and what major can get them there.
@@sdcstats Can we pursue a masters in stats after a bba ( finance accounting)? Is applied stats easier fan we be a a data analyst with just a bba i had switched from business data analytics
@@aena5995 There are a lot of Master's level statistics students who come from business, finance, accounting, or similar majors. If you're in finance accounting you probably took a course called "Econometrics" which should have some decent statistics in it. If the finance accounting courses were light on the math (linear algebra and proofs especially) you could find yourself struggling more than other students, which means certain classes (like statistical theory) will take more effort and time, but other classes (like time series) might be a little easier since you would easily see a lot of the applications. I don't know if statistics is easier or harder than business related degrees. It would depend so much on the goals and intensity of each program, as well as the skills and interest of the student themselves. It's easy to say that Master's level is going to be harder than Bachelor's level no matter what department we're talking about. Most of the students I have known which did both statistics and finance have said statistics was a little harder, but also more fun (and yes, my opinion will have bias).
The term "applied math" could just be another term for "science". Applying math can be done to do physics, or chemistry, psychology, or statistics. The idea is you have a problem, and using your expertise in that domain (whether engineering or statistics) you determine the best assumptions to frame the problem, then do math, then after the math is done you figure out how the results can be used to solve the problem you had. Computer scientists do this to figure out the best database configuration to maximize profits. Statisticians do it to solve problems in every field I've already mentioned (plus all the others) when the data has uncertainty, bias, error, and doesn't truly fit the assumptions that were used on the math. There isn't a science out there that doesn't use math, but saying that means all you need is math would be like saying all a construction worker needs is to learn how to use a drill.
@@sdcstats In that case, I think statistics is just a special case of applied mathematics. In all areas of applied mathematics, all we do is translating the observation into a purely mathematical model by making assumptions.
@@Thefare1234 Then physics, chemistry, engineering and psychology are just a special case of applied mathematics as well. But I wouldn't trust someone with a PhD in math to be my surgeon. I wouldn't say a degree in mathematics qualifies you to work at Thiokol making the space shuttle rocket fuel. And I have seen people who were brilliant at mathematics stumble when asked the most basic statistical question, like whether AIC is superior to BIC when doing regression. I had someone ask me last month to look at what demographics are not able to get internet connectivity during the pandemic and the reason why. The math part was simple. Figure out how to get the data, managing the databases, determining which model assumptions are best, interpreting what the results mean, explaining it to the client - those are all the hard parts. Especially when the answer shows that race is not a factor, but several other variables that connect to race are factors.... the math part was done in 10 seconds by the computer. The discussion is ongoing.
hi, i did math stats, anova and regression end of last year, this year i did stochastic and im currently doing time series. I want to deepen my stats knowlege any sugestions on which courses i should do next?
It might depend on what your career goals are, but if you're looking for breadth of statistical knowledge I'd recommend categorical, bayes, non-parametrics, and/or machine learning algorithms. That being said some programs build these topics into other classes instead of teaching a class solely dedicated to that one topic, so you have to see what your options are.
@@cremildamondlane3900 It depends on the school, but I was listing four classes: 1)Categorical (how to analyze data with non-numerical responses) 2) Baysian (combining data with opinion... although Baysians would find that description offensive) 3) Non-Parametrics (when it's not normal, and in fact it's not any nice distribution) 4) Machine Learning (a bucket of algorithms that use a lot of computational power - PCA, random forest, cluster analysis, etc)
@@zonerazolau-e6s It's a common debate in statistics departments to ask what level of theory is best. If you're planning on a PhD thesis that's heavily theory related then it would be helpful, especially as you're solving likelihood models with complexity (like hierarchy, bayesian analysis, or causal models). If you're doing masters level coursework it's probably less common (depending on the theory class). If it's heavily applied (instead of theoretical) then someone who doesn't know differtial equations might be just fine.
Great video. But some relationships seem really pushed here... Like English... So a Japanese who doesn't speak English can't know statistics? It doesn't feel right.
BRUNO MOREIRA GUEDES Ah you’re right of course. I was lumping “communication skills” under the English department (which is certainly the structure here) but I don’t wish to imply that English is the only language. Data Science requires good communication in some language, but any language.
Depends on your personality. People with precision and focus might find themselves drawn towards mathematics. People with broad curiosity and a keen ability to see how things move and interact will find statistics intuitive.
@@sdcstats Thank you I'm satisfied with your answer. Also if a maths major person (curious one) changes to statistics..will it be easier or harder for them to understand statistics?
@@bangtankibaraat9215 In my experience a math person can transition to statistics easier than a statistician can transition to mathematics. Statistics works because the theory behind it was developed, and a person who is good at math will get the theory well and that makes it easier to see why it all works. People who aren't good at math can still use the methods, and sometimes they are better at the consulting skills, but unless they put in a massive amount of work to get the math they'll have to memorize details that a math person would understand.
Why is statistics class so convoluted? It is so far from real life and SO, I repeat SO much information and time wasting, that it is actually difficult to learn with many of the current methods. We had to read and remember 12 chapters in week 1.......yeah it's not realistic and if I don't need all of that information, why waste my time and confuse me? Just get to the point of how to solve the problem. I feel like I am missing the psychology part and maybe that is where my frustration lies.
That's like asking which is better, a hammer or a screwdriver. Both are useful for the right project, but if you confuse them you'll just end up causing problems.
Lmao the different clothing for him imitating the student killed me
It's indeed very funny 😹
Easily one of the best video on the internet... Thanks man...
Incredible explanation. I only hope that going forward I will come across many educators like you.
This is most informative video I have ever seen on TH-cam. The level of explanation is crystal-clear.
Good video. Talks about the philosophy of statistics and delves into the meaning behind each chapter.
Thank you so much for this video. I've been passionated for stats since my first class on it, but I often question myself if I should keep pursuing this major giving my dificulteis with math. I’m definitely not the quickest with numbers but I do get the concepts. I also have a hard time memorizing facts but on the other hand I have a strong logic, so I am usually able to draw conclusions if I have time for it. It’s challenging because I see most of my classmates looking at problems and solving them within seconds and I take so long. Anyway, your explanation between the subjects is very encouraging. I’ve always loved philosophy, psychology, anthropology, and I feel like stats is the tool to put all these subjects together to better understand the world and our society so, thanks again for the encouragement.
There are a lot of statisticians out there that are lousy at math. If you know what needs to be done you can ask the computer to do the calculations. There is a branch of stastistics called "theoretical research" that is VERY math intensive, and you've gotta be good at math to succeed there, but there's all sorts of applied research and consulting options that don't require you to know all the math tricks. If you love science and get math mixed up there is a large area of statistics that is perfect for you.
@@sdcstats There are statisticians and then there are mathematical statisticians. The latter are more competent.
There are two kinds of statistics:
1. Descriptive
2. Mathematical
Anyone can learn descriptive statistics. Not anyone can learn mathematical statistics.
I am really glad I came here to understand the difference between statistics and math. The internet is filled with answers that are like the first example you used and it's nice to know it's not so simple. I am a freshman undergraduate looking at both a math and statistics major. I really like how applicable statistics is to the real world. Not that you said mathematics isn't. But maybe just not in the ways that I am looking for.
Your students are lucky! Thank you for explaining this so well 🤗
Thank you Dr. Crawford for your wonderful explanation. I especially enjoyed your breakdown of an intro stats class into math and statistical concepts. Could you please recommend an intro textbook that adequately addresses the topics you listed under statistics? Keep making videos! You are saving the world!
Well my favorite is probably "Introduction to the practice of statistics" by Moore and McCabe. The book by Agresti and Franklin "Statistics the art and science of learning from data" isn't bad either (a little more technical, less on the pictures). A lot of other teachers swear by the authors Utts (who keeps things quite simple) or Devore (who is a bit more advanced). If you want a good intro book that is advanced enough for most secondary stat classes then "The Statistical Sleuth" by Ramsey/Schafer is a good pick.
Thank you so much for your reply! You are most helpful
Brilliant educator. Concise, informative, and passionate, thank you
So good. Students should watch it in the beginning, being kind of confused, and then at the end of semester so they can click with a concept more. It's really good put together.
I wish I had professors like you while learning these subjects!
This is epic and wonderful!! Like unimaginable and well explained!! I wish to see more like this video!
Love ❤ from India sir.
"Rather than compare the two..."
-subtly includes 'fun with a thriving job market' on the stats side of the board xD
Service is the heart of statistics. I love that! It really gives me my WHY when confronted with the HOWS.
Excellent video! The fact that students who approach stats as if it were a math class do poorly is testimony to the way math is too often taught, which is sadly in a step-by-step algorithmic computational fashion, or at least that's how many students try to get through it. Instead, they should see math as rigorized thinking and precise expression of relevant philosophical concepts such as form, pattern, distance, closeness, continuity, etc. The "step by step procedures" in textbooks is partly responsible for this. So they learn to manipulate formulas, etc., using this approach, pass the course, then get to stats and think they can do the same to succeed. But that kind of thinking doesn't work in stats (assuming the course is taught well), because of all the reasons Scott mentions. Stats, when combined with research and science, requires many inputs to reasoning through a problem, and in scientific applications, it can prove very difficult figuring out optimal designs and appropriate analyses. Not only can it be difficult verifying assumptions, etc., but even determining whether your variables are best measured on a pseudo continuous scale or categorical can be very difficult. And, even once you've conducted your analyses and answered your hypotheses, determining what you can vs. cannot conclude scientifically can be very challenging. Most students greatly overestimate what can be concluded from a statistical model. What is usually more relevant to drawing strong scientific conclusions is the underlying research design. As Scott discusses, it's never as simple as simply running a statistical test. When you're working with statistics, you're usually working with science at the same time, and all the things Scott mentions has to be considered in your decision-making and in drawing conclusions from your analytical method. You have to be well-versed in the field in which you are applying the statistics so you know the types of conclusions that can vs. cannot be drawn on a substantive level.
This guy is extraordinarily good at explaining things. He makes it look so simple!
Fantastic explanation. Superb editing. Transition shots taken so meticulously that the edit does not seem jump cut (except few places). The video making helps the topic to be more interesting. Thanks for the video!
I have never seen anyone on YT explain subjects so well . Also , thank you millions for not playing music in the background
Man this is the best explanation I have ever seen. Thanks for sharing!
Nice video. I love how you explain. simple and creative
The best editing of a video ever!
I love how you explain everything... I have a question.
I am planning to get my masters degree in statistics science... however I suck at math.
but im good in marketing and understanding what people are looking for and why through patterns in habits... do you think this is a right path for me?
please explain
There are a lot of very successful good statisticians out there who aren't very good at math. Thanks to technology you absolutely can become a great data scientist without great math skills.
That being said - there is at least a certain amount of math that is expected: for example if you understand how an integral works then you'll be more comfortable with what the computer is doing when you get probabilities under some distribution. You've got to be comfortble with fractions, and how to read a polynomial equation. But you don't need much beyond that.
As for getting a degree without great math skills - now that depends on the program. You'd want to ask questions like "Is this program heavily theoretical or more applied?" The more theoretical it is the more expectations they will have of good math skills. Still no matter where you go there's going to be a theory class, and you'd have to struggle through it. The good news is that after fighting through it if you feel like your theory skills are still weak you can still be a great statistician.
Only if we had more teachers like this guy.
Need to further break it down to inferential and descriptive statistics.
The explanation was splendid!
Underrated channel 🫠
Thank you for this video Mr.Crawford.
I was so confused.. thanks to you for this amazing explanation!
Thank you for making this video!
Damn top notch explanation.
Thank you. This really helped me out.
Good information thank you Love from india ❤️
You are a rock star!
Amazing explanation. Very much needed to watch this 🤞
Brilliant, just the stuff we were looking for - thank you for your service!
... Unfortunately, a lot of the elderly people +50 have no idea what they are doing, while living ... so we have to teach the following generation. And a lot of the elderly people in the deep state seem to be just cowards, they collect the data on the general public - and, with some help of some younger guys, they know what they are doing, which is exactly what you showed to us. Just applied on the general public, all-time, everywhere. And this is why they are cowards with fear to loose control: A public being as smart and as well informed as they are would help to develop the societys forward. There is no use for secrets - cause our society's have to use their brains to the max. Maybe together with a good cup of tea from time to time, yes.
Mathematical Statistics is no joke. IMHO, one of the toughest module around.
A little bit reductive on the mathematics side but a very good explanation of how statistics is done!
dude, stats is way bloody harder than raw calculus, more specifically, probability. probabilities suck. it's bloody interesting, but it's bloody hard
Excellent detailed explanation, thank you for the video
Great video! Great orator skills. Great video editing skills. Great teaching...order of information, explaining a new concept by referencing/leveraging concepts that the student already knows and building off of it.
GRRRR! Why can't all teachers be this good. Can we just let STEM individuals restructure everything in society! Get rid of incompetent narcissists who can't innovate or solve a problem to save their life so they instead paywall things like natural resources and education and then go on to artificially create self-worth by creating and perpetuate myths like "engineers lack social skills."
Brilliant video! Thanks so much!
"...because statisticians are the BOSS." LOLL
Wonderful explanation and overview!
'It's not the equations that trip people up'
Thank you for this.
He is funny. Also great lecture. Thanks a lot
Statistics become more promising to me after I watch this video:)
Thank you. This is very helpful.
I'm doing a mathematics degree in UK and theres a unit for mathematical statistics and it's VERY mathsy and doesn't deal with much raw data an anlaysis. But I guess this is more the theory behind it so maybe thats why
"[Statistics] does include some math. Not really a whole lot (...)" ... still too much for most :D
This was amazing.
I got 2 words for ya .... “SET THEORY!!🔘🔘
Honestly, I feel like learning statistics would connect some of the dots I feel were missing from my thermodynamics class...
After thermodynamics you learn statistical mechanics
We are using different sciences to solve world problems. Statistics is a science to add improvements where there is no permanent solution. Math is a science which can resolve problems permanently and their should not be any room for improvements.
Do I need to understand differential equations to understand time series or stochastic modeling?
Awsome vídeo.
Great video!
What does an arithmetic mean tell one?
[A] Nothing
[B] It's just an average
[C] It's easy to compute: just add up the numbers and divide by the total
[D] The arithmetic mean is only valid when data can be redistributed within a given set, meaning that in such cases, there are never any outliers.
[E] Class arithmetic means are useless because students cannot share their grades
Which is the best description of what is an arithmetic mean?
I wanted to comment on your other response, but I don't see it anymore. I've had people say their comments got lost, and I don't know why. You made a good point that knowing mathematics is going to be a key ingredient to high level statistics. I made this video because the place where I used to work was combining the mathematics and statistics departments, and many of the math faculty couldn't understand how we were different. They really believed we just taught "applied math".
It was a big deal when it came time to evaluate my job performance because they were so different. When I showed them how many research projects I had been associated with they accused me of "moonlighting". I made this video in part to help them understand how our statistics classes were very different than "applied algebra". It was a wild time, and I'm glad I moved back into a separated statistics department.
Your question above is a fun one - each answer could be valid depending on what you're trying to do. There certainly are examples where the arithmatic mean isn't useful, but others where it very much would be. Class means always drive me nuts - students will take a test and ask "what was the average" and I cringe thinking "It was part of the exam to ask why you wouldn't want the average in this case...." In engineering the mean is called the "first moment" because you're collapsing the entire data set into one value. In physics they call it the center of gravity. Knowing the center of gravity is just as useful as knowing the mean. There's scenarios where it would not be sufficient, but depending on the question you're asking it might be a great way to simplify the calculations. If I really had to pick an answer I think I'd go with C.
@@sdcstats The correct answer would be [D] because that's exactly what happens when you do that which is described in [C], ie, you redistribute the values of each datum so that they are all equal.
An arithmetic mean is useful when it makes sense to redistribute the values so they are all equal. On the other hand, it makes no sense to compute an arithmetic mean when it makes no sense to redistribute the values.
For example, an arithmetic mean of peoples' height is useless, because you can't chop them all up so that they all have the same height. This is the same as taking a class grade - it tells you nothing about a student's performance. In a class of 10 students, the following scores are recorded:
30,30,30,30,30,30,30,90,100,100
The mean is 50 implying that most students would have passed, but this is obviously FALSE. 7 students failed.
Now for a useful mean: Suppose there are 3 cities that each require 2 million cubic meters of water each month. If we are at the end of a month, how can we tell if the current water reservoirs have sufficient water for the next month? That's right! We determine the arithmetic mean of the three cities reservoirs that are connect by conduits. If the mean is 2 or greater then we are done. Otherwise, we will have a water shortage.
City A has 1 million cubic meters
City B has 2 million cubic meters
City C has 3 million cubic meters
In the above scenario, the mean tells us that we have enough water if it is redistributed or shared between the cities.
In fact the mean value theorem which is the reason calculus works, is about an arithmetic mean.
Make sense?
@@sdcstats What I am telling you is that you can know beforehand if it is logical or not to calculate an arithmetic mean. You can know.
If redistribution of values makes sense, then the arithmetic mean will be valid with no outliers or anything else unexpected.
If redistribution of values does not makes sense, then no need to calculate an arithmetic mean.
@@RealMathematician21stCentury Cool. Thank you for the insight.
@@sdcstats You're welcome!
I'm weak in maths my choice is i want to go statistics what to do sir?
You don't need to be good at maths to do well in statistics. You can leverage your math skill with computer science skill (basically asking the computer to do the math for you) or a business analyst direction (knowing how to be part of a team where some do the math and you do the planning). Not wanting the heavy math does limit some options (like a theoretical PhD) and not every school can offer a direction that has less math, but those options certainly do exist and can lead to happy careers.
I might major in Statistics. It sucks that I need a Master's degree to be considered competent. Like what? Four years isn't enough to cram all those math classes?
There are some schools (like A&M) where a Bachelors in Statistics means the student is suprisingly competent. But many schools treat the bachelors in statistics as a math degree with some extra probability and basic statistics vocabulary. Because of that businesses often view a bachelors in statistics as "They know how to handle data, and they're good with complex instructions, so they can be number crunchers." On the other hand masters students are seen as "They understand statistical methods and have some experience with real application of the common types of analysis", so they do the analysis after the bacehlor has cleaned the data. Then you have PhD expectations: "They either know all the methods, or can figure them out with a little bit of time. They could go beyond what's been done and do something new or unique if we needed.", so they meet with the CEO to talk about what is needed, plan how to get the data, and tell the master's level hire what analysis to do after the bachelor has cleaned the data.
Well done.
bro you are best
I have a question sir
I'm about to start my journey studying statistics but I'm bad at maths to be honest
Please what is your advice to me?
1) No one is "bad" at maths. It just means it will take you longer to learn than others, but it takes everyone time, and if you're slower you CAN still learn it.
2) Find good friends. Please who do see mathematical things quicker can help you get throught the tougher things. Don't be a leech, no one wants someone who pretends to be a friend just to get answers. But be kind and friendly to others, work hard, and they will recognize you are a good person and help teach you.
3) Aim towards classes that are more applied and less theoretical. That depends on what school you go to, but often classes like regression, categorical, nonparametric, and data mining will emphasize the applied side. Often classes like multivariate, time series, spatial, bayes (advanced) will be more theoretical. That school dependent, though, so ask around. You'll also find certain teachers tend to emphasize the theory/application more, so gravitate towards the teachers that follow your style.
4) Be patient. There is no shame in repeating a class, especially if it's a theory heavy course. You can make it, and you can be a good statistician without good maths skills. If it takes you longer to graduate no one will care 3 years later. Always have the attitude of "Wow, this course is hard, but I can do it!"
@@sdcstats wow
Thank you so much sir
I really appreciate you taking time to explain this to me. with what you have said I believe I can do it 💪🏼
Well there is Mathematical Statistics which, when I took it had a lot of math and probability theory.
You are good.
Math and stats which one is better please 🙏🙏🙏
It will depend on your skills and your career goals. If you're a very systematic formulaic thinker you'll probably like math better. If you're the type to see the big picture and want to do concrete applied things then you'll like statistics better. If you're planning on graduate schoool those programs (pretty much all of them) like math majors, because they can focus on teaching their subject instead of the math tricks they use. Statistics will teach you their subject while talking about all the others at the same time (engineering examples, sociology examples, biology examples, etc.)
If you want to be in academia (a teacher or a professor) then you can pick whichever subject interests you most. If not, then math should be a jumping off place where you eventually move into the subject you're interested in (towards your career). Statistics by itself leads into lots of great careers that are in demand now and expected to stay in demand for a long time. You wouldn't have to switch majors/careers later, and you can steer towards the type of statistics that deals with the field of science you're most interested in.
thanks
Hello I want to be data scientist and machine learning engineer should I go for math or statistics?
Well my opinion will ovbiously be biased towards statistics, but it does sort of depend on where your goals are. If you want to be a theoretical researcher or go for a PhD at a top tier research institution go for math. If you're aiming for a business, medical, financial, or commercial occupation then you'll want statistics. If you're aiming for cyber security or database management or big data machine learning algorithms then you might prefer computer science. And all of these jobs have shades of grey and overlap.
What many undergraduate students do is pick the major (math, statistics, computer science, business, or econ) that they enjoy and are good at, and then after 4 years of focused study they have a better idea of what career path they are interested in and what major can get them there.
@@sdcstats
Can we pursue a masters in stats after a bba ( finance accounting)? Is applied stats easier fan we be a a data analyst with just a bba i had switched from business data analytics
@@aena5995 There are a lot of Master's level statistics students who come from business, finance, accounting, or similar majors. If you're in finance accounting you probably took a course called "Econometrics" which should have some decent statistics in it.
If the finance accounting courses were light on the math (linear algebra and proofs especially) you could find yourself struggling more than other students, which means certain classes (like statistical theory) will take more effort and time, but other classes (like time series) might be a little easier since you would easily see a lot of the applications.
I don't know if statistics is easier or harder than business related degrees. It would depend so much on the goals and intensity of each program, as well as the skills and interest of the student themselves. It's easy to say that Master's level is going to be harder than Bachelor's level no matter what department we're talking about. Most of the students I have known which did both statistics and finance have said statistics was a little harder, but also more fun (and yes, my opinion will have bias).
Thank you
I still don't get the difference. What you explained is just like applied math. As soon as the assumptions are made it turns into normal math.
The term "applied math" could just be another term for "science". Applying math can be done to do physics, or chemistry, psychology, or statistics. The idea is you have a problem, and using your expertise in that domain (whether engineering or statistics) you determine the best assumptions to frame the problem, then do math, then after the math is done you figure out how the results can be used to solve the problem you had. Computer scientists do this to figure out the best database configuration to maximize profits. Statisticians do it to solve problems in every field I've already mentioned (plus all the others) when the data has uncertainty, bias, error, and doesn't truly fit the assumptions that were used on the math. There isn't a science out there that doesn't use math, but saying that means all you need is math would be like saying all a construction worker needs is to learn how to use a drill.
@@sdcstats In that case, I think statistics is just a special case of applied mathematics. In all areas of applied mathematics, all we do is translating the observation into a purely mathematical model by making assumptions.
@@Thefare1234 Then physics, chemistry, engineering and psychology are just a special case of applied mathematics as well. But I wouldn't trust someone with a PhD in math to be my surgeon. I wouldn't say a degree in mathematics qualifies you to work at Thiokol making the space shuttle rocket fuel. And I have seen people who were brilliant at mathematics stumble when asked the most basic statistical question, like whether AIC is superior to BIC when doing regression.
I had someone ask me last month to look at what demographics are not able to get internet connectivity during the pandemic and the reason why. The math part was simple. Figure out how to get the data, managing the databases, determining which model assumptions are best, interpreting what the results mean, explaining it to the client - those are all the hard parts. Especially when the answer shows that race is not a factor, but several other variables that connect to race are factors.... the math part was done in 10 seconds by the computer. The discussion is ongoing.
hi, i did math stats, anova and regression end of last year, this year i did stochastic and im currently doing time series. I want to deepen my stats knowlege any sugestions on which courses i should do next?
It might depend on what your career goals are, but if you're looking for breadth of statistical knowledge I'd recommend categorical, bayes, non-parametrics, and/or machine learning algorithms. That being said some programs build these topics into other classes instead of teaching a class solely dedicated to that one topic, so you have to see what your options are.
@@sdcstats thank you, just to be clear, "categorical" and "bayes non-parametrics" can be seen as 2 courses right?
@@cremildamondlane3900 It depends on the school, but I was listing four classes:
1)Categorical (how to analyze data with non-numerical responses)
2) Baysian (combining data with opinion... although Baysians would find that description offensive)
3) Non-Parametrics (when it's not normal, and in fact it's not any nice distribution)
4) Machine Learning (a bucket of algorithms that use a lot of computational power - PCA, random forest, cluster analysis, etc)
@@sdcstats what do you think, sir, do I need to understand differential equations to understand time series or stochastic modeling?
@@zonerazolau-e6s It's a common debate in statistics departments to ask what level of theory is best. If you're planning on a PhD thesis that's heavily theory related then it would be helpful, especially as you're solving likelihood models with complexity (like hierarchy, bayesian analysis, or causal models). If you're doing masters level coursework it's probably less common (depending on the theory class). If it's heavily applied (instead of theoretical) then someone who doesn't know differtial equations might be just fine.
Great video. But some relationships seem really pushed here... Like English... So a Japanese who doesn't speak English can't know statistics? It doesn't feel right.
BRUNO MOREIRA GUEDES Ah you’re right of course. I was lumping “communication skills” under the English department (which is certainly the structure here) but I don’t wish to imply that English is the only language. Data Science requires good communication in some language, but any language.
I can't tell whether this guy is 40 or 17
Which one is hard to understand?
Depends on your personality. People with precision and focus might find themselves drawn towards mathematics. People with broad curiosity and a keen ability to see how things move and interact will find statistics intuitive.
@@sdcstats Thank you I'm satisfied with your answer. Also if a maths major person (curious one) changes to statistics..will it be easier or harder for them to understand statistics?
@@bangtankibaraat9215 In my experience a math person can transition to statistics easier than a statistician can transition to mathematics. Statistics works because the theory behind it was developed, and a person who is good at math will get the theory well and that makes it easier to see why it all works. People who aren't good at math can still use the methods, and sometimes they are better at the consulting skills, but unless they put in a massive amount of work to get the math they'll have to memorize details that a math person would understand.
@@sdcstats Thank you! You explained it very well.
stats here I come
BOSS😂🔥
nice
Genius
Why is statistics class so convoluted? It is so far from real life and SO, I repeat SO much information and time wasting, that it is actually difficult to learn with many of the current methods. We had to read and remember 12 chapters in week 1.......yeah it's not realistic and if I don't need all of that information, why waste my time and confuse me? Just get to the point of how to solve the problem.
I feel like I am missing the psychology part and maybe that is where my frustration lies.
Try and figure out a p value and you might start an argument. Spss was made for us dummies who are Jack of all trades but masters of none.
Forgot to say, that's why the news can trick everyone all the time...because what, you don't believe in science?
which is better
That's like asking which is better, a hammer or a screwdriver. Both are useful for the right project, but if you confuse them you'll just end up causing problems.
Math of course
Both
Statistics is art. Math is fact
soa KoHaji statistics are literally a set of facts but okay lmao
you need english for math tho... especially the pure stuff ;)
i see your point though!
Hell yea