You are a gifted teacher. I am so grateful for your presence on RUclips. I feel I've learned so much more from you than I have from my text. I cannot thank you enough!
Your videos and teaching style are SOOOOO critical. Many of us struggle with the non-intuitive formulas and have given up. This stuff CANNOT be applied to real life if we cannot conceptualize what they really mean at their core and why they were derived in the first place. You are bridging that gap Justin, unlike so many others before you, and I am forever grateful. Thank you.
You're a wonderful teacher. Covariance and correlation was a mystery to me before. But you not only explain them in a PDF but also compare with a sample, which makes it extra clear to understand. Thank you so much !!!
Absolutely brilliant presentation! As a (high school/ secondary school) teacher of these topics, I find that you ALWAYS help me clarify my own explanations, making this difficult stuff much easier for students to grasp - I especially agree that overuse of formulae is dangerous: students who overuse formulae without a basic understanding of the processes risk failing to interpret results properly; this problem gets worse NOT better, the more more they do with statistics without such basic understanding! Well done, Justin, and many thanks for all your hard work!
@@zedstatistics Sadly, applying formulas is precisely what most schools teach kids to do in maths classes... maybe not anymore nowadays when lots of schools have incorporated CS into their curriculum? But during my time in elementary up until high school I had been taught only the formulas..urrrgh
Good work. I have never watched even one video on RUclips explaining that how statistical formulas are constructed. We just apply them but never learn and taught how to make them instead.
Brooooooooooo it's safe to say teaching is your calling. You're gifted at teaching and your explanations are really top notch. Please i'm struggling with Simple linear Regression and OLS and i'll need your help with that
It's great to have you! I'm having an exam in statistics soon, and they are going to test us mainly on the theory. It was so hard to find good videos that explain the logic behind everything and not just how to plug numbers in a formula. Many thanks
Can’t appreciate more about your beautiful works. I am studying machine learning, but struggling with some statistics concepts . Your videos really help me out !! You put these knowledge in a structural way, which is really impressive!!
Thank you soooooo muchhhhhh. I've been struggling with statistics for years. Your videos helped me understand much more than my university teachers although English is not my mother tongue.
Your presentation skills are great. The visuals and the Excel exercise displays are appreciated. The text/content I have is not as "clear" as your teaching approach via RUclips. Many thanks to you!
Hi Justin I watched the entire descriptive statistics videos series. I must say all the videos are simply amazing in every imaginable way. I have zero background in statistics and i enjoyed all the videos tremendously. thank you for taking the time to make them as it inspired me to dig deeper into the statistics field. Hope this message will reach you and inspire you to keep making this type of great content.
Here is the gist: Covariance gives you one piece of information. It measures the direction of the two variables (positive or negative) positive (direct relationship) means that when one variable increases the other increases and negative means that when one variable increases the other decreases (inverse relationship), whereas, correlation gives you two pieces of information. It gives you the direction (positive or negative) and the strength of the relationship by being a number between (0 and 1). If two variables have a very strong relationship their coefficient might be .90 and if there is a weak relationship the coefficient might be .20
Thank you so, so much for these videos. They are immensely helpful in actually understanding statistics as opposed to just memorizing things. I’m learning statistics for work and your videos make things so interesting and easy to comprehend. Truly amazing explanations!
I'm finishing a Data Science speciaization on Coursera and the current course is dry as hell, feels like I'm eating an yogurt made of sand. This video is an oasis ou clarity and ease. I'm gonna watch all your videos on the same topics of the other course, this one helped me immensely. Thanks a lot!
He uses the sample standard deviation, which in numpy can be obtained by passing a second parameter 'ddof=1', like: numpy.std([30, 35, 40, 25, 35], ddof=1)
In summary, when you're working with a probability distribution, the probabilities already adjust the contribution of each outcome to the covariance, so there's no need to further divide by n or n - 1
Hi! Your series about the regression was perfectly done and it really helps understand the concept. Could you please do the similar one in topic: all about the ANOVA method?
Nice video and series. Do you have a tutorial on using the covariance matrix to estimate the variance of a sum or product? I recently found out how to do this and realized that people do not cover this very often.
Really enjoying these videos, Justin. Please keep 'em coming. Although I understand the distinction between correlation and interaction, I'd be interested to see how you might explain it in your inimitable fashion.
Yo. This is what I have been trying to do to gain a better understanding. I would like to see the derivation of all PDFS from scratch . That would be amazing, but it seems like even P.H.D students have a hard time with this for some reason and I havent found it online. Usually the proofs for these things are done through moment generating functions. I believe you would have to start first with chi-quared, then go to something like t because to involves chi-squared.
Thanks for the video. If the mean is the average of weighted X, to find the deviation, the X should also be multiplied by the probability weights right?
please please do a complete tutorial on panel data analysis!!!!im starting my phd and i feel safe to know that you taking chrge of my model!!!will you???if you do please hurry!!ive searched today in youtube about panel data and there is a huge gap between you and the others!!!!
You are very good at building intuition, something my university teachers were very lacking. Thank you for this video and your efforts. One question that always bugs me and can't really figure it out is that for me it seems like this kind of thing has to do with the order in which you put the values in the table. I'm more concerned with the theoretical model. Like It would gave the same results if you just shuffled the rows around, but then it wouldn't give the intuition (or not as obvious) of seeing how they are negatively correlated. Like I'm thinking let's say we have a two samples x, y, you can order them of course in whatever way you want cause they were generated or measured randomly (eg. asking people in different order). When I'm thinking graphically, I always see putting the x values in ascending order, but how you draw the y values depends on what order you are taking them, right? And in which order you draw them could can cause positive or negative correlation or none as well. I think I'm kind of confused : )
You are a gifted teacher. I am so grateful for your presence on RUclips. I feel I've learned so much more from you than I have from my text. I cannot thank you enough!
Your videos and teaching style are SOOOOO critical. Many of us struggle with the non-intuitive formulas and have given up. This stuff CANNOT be applied to real life if we cannot conceptualize what they really mean at their core and why they were derived in the first place. You are bridging that gap Justin, unlike so many others before you, and I am forever grateful. Thank you.
Thanks!
Your explanations are so impressive. I just know I can't forget the concepts ever again
You're a wonderful teacher. Covariance and correlation was a mystery to me before. But you not only explain them in a PDF but also compare with a sample, which makes it extra clear to understand. Thank you so much !!!
Absolutely brilliant presentation! As a (high school/ secondary school) teacher of these topics, I find that you ALWAYS help me clarify my own explanations, making this difficult stuff much easier for students to grasp - I especially agree that overuse of formulae is dangerous: students who overuse formulae without a basic understanding of the processes risk failing to interpret results properly; this problem gets worse NOT better, the more more they do with statistics without such basic understanding! Well done, Justin, and many thanks for all your hard work!
Love how you always try to build a strong intuitive understanding of formulae!
Thanks, SS! Ain't no point teaching people to simply apply formula - that's what you get computers to do :)
@@zedstatistics Sadly, applying formulas is precisely what most schools teach kids to do in maths classes... maybe not anymore nowadays when lots of schools have incorporated CS into their curriculum? But during my time in elementary up until high school I had been taught only the formulas..urrrgh
Good work. I have never watched even one video on RUclips explaining that how statistical formulas are constructed. We just apply them but never learn and taught how to make them instead.
Brooooooooooo it's safe to say teaching is your calling. You're gifted at teaching and your explanations are really top notch. Please i'm struggling with Simple linear Regression and OLS and i'll need your help with that
I am from Asia , and I love you man , Your explanation is so great ‘ easy to understand
This is a stupendous explanation. You've taught us array of concepts in a single video (especially the intuition behind df).
It's great to have you!
I'm having an exam in statistics soon, and they are going to test us mainly on the theory. It was so hard to find good videos that explain the logic behind everything and not just how to plug numbers in a formula.
Many thanks
Can’t appreciate more about your beautiful works. I am studying machine learning, but struggling with some statistics concepts . Your videos really help me out !! You put these knowledge in a structural way, which is really impressive!!
Thank you soooooo muchhhhhh. I've been struggling with statistics for years. Your videos helped me understand much more than my university teachers although English is not my mother tongue.
Your presentation skills are great. The visuals and the Excel exercise displays are appreciated. The text/content I have is not as "clear" as your teaching approach via RUclips. Many thanks to you!
Hi Justin
I watched the entire descriptive statistics videos series. I must say all the videos are simply amazing in every imaginable way. I have zero background in statistics and i enjoyed all the videos tremendously.
thank you for taking the time to make them as it inspired me to dig deeper into the statistics field.
Hope this message will reach you and inspire you to keep making this type of great content.
Here is the gist: Covariance gives you one piece of information. It measures the direction of the two variables (positive or negative) positive (direct relationship) means that when one variable increases the other increases and negative means that when one variable increases the other decreases (inverse relationship), whereas, correlation gives you two pieces of information. It gives you the direction (positive or negative) and the strength of the relationship by being a number between (0 and 1). If two variables have a very strong relationship their coefficient might be .90 and if there is a weak relationship the coefficient might be .20
Correlation is non negative? So correlation doesn't give direction
Thank you so, so much for these videos. They are immensely helpful in actually understanding statistics as opposed to just memorizing things. I’m learning statistics for work and your videos make things so interesting and easy to comprehend. Truly amazing explanations!
please make more stats vids ! My lecturers just read off slides and doesn't explain the work. These are excellent !!
I'm finishing a Data Science speciaization on Coursera and the current course is dry as hell, feels like I'm eating an yogurt made of sand. This video is an oasis ou clarity and ease. I'm gonna watch all your videos on the same topics of the other course, this one helped me immensely. Thanks a lot!
Fantastic explanation!! Very intuitive and easy to understand .
Sir, you're the best at this. Hands down.
best stats teacher on youtube, thank you so much for your videos here
Wish I had this when I was struggling through in college. Excellent video as always!
I bow down to your knowledge sir, Namastey!
You explained better than my lecturer and textbook! Thank you.
What a boon for people new to statistics .Thank you for explaining so logically !:)
You are a gift to the world!
Your videos greatly aid understanding of a concept. Thank you for your contribution to society!
Another great video explanation, reminds me of my A Level stats many moons ago!
8:23 how did you get this values of 5.70 and 1.87. I try to understand the calculation. It isn't the variance or standard deviation.. plz help
He uses the sample standard deviation, which in numpy can be obtained by passing a second parameter 'ddof=1', like: numpy.std([30, 35, 40, 25, 35], ddof=1)
x = -3 +2 +7 -8 +2
square(x) = 9 4 49 64 4
sum(square) = 130
divide by n - 1 = 130/4 = 32.5 [variance]
srqt(variance) = 5.70 [standard deviation]
same for y column : )
@@85aayush5 THX (:
Zedstatistics. You have out done yourself!!!👏👏👏
Absolute Gold - this series
So well put together and crystal clear! Thank you!
great stat teacher on yt
Thank you so so so much! I thank God for you. Keep doing what you are doing.
I love you man... You saved me 😭❤️
Why are statisticians interested in covariance if it is not informative as correlation? EXCELLENT VIDEO.
Thank you for explaining this all easily!!
Simply amazing!! Never thought it's so easy!! 🙏
Just wanted to say these are great videos and you are doing an amazing job producing these. Please keep them up!
this channel is a blessing!
Thank you very much Sir, one of the best explanation on the planet. From Iran
A gifted teacher
Bless your heart, Justin Z.
You are an awesome teacher!!!! Thanks.
Great explanation, I love the quick and dirty explanations.
Brilliant description on covariance
This is such great prep before doing a business stats unit for the first time. Thanks heaps Justin!!
I just loved your explanations.it was very helpful
I should give YOU my tuition. Thank you sir!
You're perfect teacher
Such a clear and logical explanation! Thanks a lot man!
I love your videos. Made my concepts clear. Please make videos of design of experiments. All the deisgns . I'm suffering so much in my studies
Thank you for this useful content my friend.
Thank you so much for this explanation!!!!
Thank you, it was extremely informative 😁. You take your time and explain it very clearly.
perfect explanation👍
Thank you lots for the awesome video that actually focused on what can be accomplished by using the formulas and what they cosist of.
Very useful, Thank you so much ❤️
why didn't you divide (n-1) at 15:33 ?
In summary, when you're working with a probability distribution, the probabilities already adjust the contribution of each outcome to the covariance, so there's no need to further divide by n or n - 1
Very good, thank you Justin.
Amazing video! Thank you.
Hi! Your series about the regression was perfectly done and it really helps understand the concept. Could you please do the similar one in topic: all about the ANOVA method?
Check out his video titled: "Regression output Explained", he talks about ANOVA there.
Love from India ❤️
Nice video and series. Do you have a tutorial on using the covariance matrix to estimate the variance of a sum or product? I recently found out how to do this and realized that people do not cover this very often.
Thank you so much clean explanation
Really enjoying these videos, Justin. Please keep 'em coming. Although I understand the distinction between correlation and interaction, I'd be interested to see how you might explain it in your inimitable fashion.
Time series please
Can someone explain how he got 5.70 and 1.87 at 8:25?
Thanks in advance!
It's the sample standard deviation of x, and std dev of y. On excel, it's:
=stdev.s("x values") = 5.70
=stdev.s("y values") = 1.87
@@KenSyRacing when you calculate it , it's actually 5,15 and 1,67 tho
THE CONTENT I NEED- SUBSCRIBED
First!
* Could you please make videos about Time series...?
Been asked about this a few times, actually! Safe to say i should probably get onto it hey!
zedstatistics Love you 🥺 best stats Channel in the history of RUclips😭😭
@@zedstatistics Hi , Zed , why do we divide covariance by standard deviations, to get correlation. Can you please tell me ?
@@zedstatistics It would be great!
@@zedstatistics I love you too, no homo... scedasticity
Thank you so much. It was very helpful
Excellent video! Did u happen to cover proof/intuition behind why correlation lies between -1 and 1? Thanks
Блестяще. Как раз мой уровень))
Very clear.. Amazing !!
Great videos. It would be nice with some time series :-)
Great lesson. Thank you!
Tnx alot sir. Nice explaining
Yo. This is what I have been trying to do to gain a better understanding. I would like to see the derivation of all PDFS from scratch . That would be amazing, but it seems like even P.H.D students have a hard time with this for some reason and I havent found it online. Usually the proofs for these things are done through moment generating functions. I believe you would have to start first with chi-quared, then go to something like t because to involves chi-squared.
While finding correlation between x & y , we are dividing cov(x,y) by the product of sd(x) and sd(y).What are we actually doing here?How does it work?
Thank you so much for the videos. They are truly helpful.
Can you make videos for time series and panel data analysis?
Hey Justin, Can you make some explanatory videos about panel data, fixed effects please? Thank you very much! Love your content!
Please make videos on statics in deep learning
Thank you for all 🤍
THANK YOU FOR EVERYTHING
Thanks for the video. If the mean is the average of weighted X, to find the deviation, the X should also be multiplied by the probability weights right?
Could you please elaborate on why did you use the joint probability of X and Y when calculating the Variance of X?
Marvelous ❤
Love it! you're amazing, keep going!
oksh ya noor 😂😂😂
Just a boss public bro🥰
Hi zed.
You forgot to explain the intuition behind why we divide covariance by the 2 std dev to get the correlation.
Could you explain it?
Hey. Can u make a video on step by step course for series of video on stats, to be consumed in your tutorial.
Thank you, just thank you.
Can you please tell what is mean by interclass correlation coefficient
how do you get the (5.70) and the (1.87) in the CORR formula?
Can you do a series on sampling distribution, pleasee!
please please do a complete tutorial on panel data analysis!!!!im starting my phd and i feel safe to know that you taking chrge of my model!!!will you???if you do please hurry!!ive searched today in youtube about panel data and there is a huge gap between you and the others!!!!
You are very good at building intuition, something my university teachers were very lacking. Thank you for this video and your efforts.
One question that always bugs me and can't really figure it out is that for me it seems like this kind of thing has to do with the order in which you put the values in the table. I'm more concerned with the theoretical model. Like It would gave the same results if you just shuffled the rows around, but then it wouldn't give the intuition (or not as obvious) of seeing how they are negatively correlated. Like I'm thinking let's say we have a two samples x, y, you can order them of course in whatever way you want cause they were generated or measured randomly (eg. asking people in different order). When I'm thinking graphically, I always see putting the x values in ascending order, but how you draw the y values depends on what order you are taking them, right? And in which order you draw them could can cause positive or negative correlation or none as well. I think I'm kind of confused : )
can someone do SD of x and y in correlation, I am getting different figures?