What is COVARIANCE? What is CORRELATION? Detailed video!

Поделиться
HTML-код
  • Опубликовано: 29 сен 2024

Комментарии • 145

  • @theyreeatingthedoggs
    @theyreeatingthedoggs 4 года назад +103

    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!

  • @economicskibaat8013
    @economicskibaat8013 4 года назад +38

    Your explanations are so impressive. I just know I can't forget the concepts ever again

  • @Maihuit
    @Maihuit 4 года назад +16

    First!
    * Could you please make videos about Time series...?

    • @zedstatistics
      @zedstatistics  4 года назад +17

      Been asked about this a few times, actually! Safe to say i should probably get onto it hey!

    • @Maihuit
      @Maihuit 4 года назад +6

      zedstatistics Love you 🥺 best stats Channel in the history of RUclips😭😭

    • @ramakanthrama8578
      @ramakanthrama8578 4 года назад +2

      @@zedstatistics Hi , Zed , why do we divide covariance by standard deviations, to get correlation. Can you please tell me ?

    • @tonimigliato2350
      @tonimigliato2350 4 года назад

      @@zedstatistics It would be great!

    • @Shervind
      @Shervind 4 года назад

      @@zedstatistics I love you too, no homo... scedasticity

  • @beasthunt
    @beasthunt 2 года назад +1

    This is great, but you did not explain what the significance of what a covariance of 8.75 is. How is this basic concept missed with you guys. Is it because everyone struggles with math so you are focused on math and not the underlying meaning of the words? All we've learned is that it's a positive correlation and that really means nothing.

  • @geoffbakerhytch5623
    @geoffbakerhytch5623 2 года назад +5

    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!

  • @Vyshu5
    @Vyshu5 2 года назад +2

    You haven’t answered why do we divide covariance by standard deviations to get correlation value🥲 I hope you can answer me.. help me Justin🥺

  • @wanderlisi5038
    @wanderlisi5038 3 года назад +5

    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

    • @joaoborges2014
      @joaoborges2014 3 года назад

      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)

    • @85aayush5
      @85aayush5 3 года назад +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 : )

    • @nadiab741
      @nadiab741 3 года назад +1

      @@85aayush5 THX (:

  • @nildev7264
    @nildev7264 3 года назад +6

    This is a stupendous explanation. You've taught us array of concepts in a single video (especially the intuition behind df).

  • @buildingourdaydream9804
    @buildingourdaydream9804 2 года назад +4

    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.

  • @liamhoward2208
    @liamhoward2208 3 года назад +4

    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

    • @sathish297
      @sathish297 4 месяца назад +1

      Correlation is non negative? So correlation doesn't give direction

  • @md.shahabulalam9484
    @md.shahabulalam9484 4 года назад +4

    Time series please

  • @98ujaal
    @98ujaal 4 года назад +22

    Love how you always try to build a strong intuitive understanding of formulae!

    • @zedstatistics
      @zedstatistics  4 года назад +14

      Thanks, SS! Ain't no point teaching people to simply apply formula - that's what you get computers to do :)

    • @xXxIMMORTALxXx
      @xXxIMMORTALxXx 4 года назад +1

      @@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

  • @amirmelloul199
    @amirmelloul199 4 года назад +17

    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.

  • @Karen-2023
    @Karen-2023 3 года назад +4

    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!

  • @athanasiosklimentidis6975
    @athanasiosklimentidis6975 4 года назад +2

    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?

  • @rezaerabbi2492
    @rezaerabbi2492 3 года назад +1

    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?

  • @howardlau7749
    @howardlau7749 2 года назад +3

    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!!

  • @dominiclederer9564
    @dominiclederer9564 5 месяцев назад +1

    please make more stats vids ! My lecturers just read off slides and doesn't explain the work. These are excellent !!

  • @meilinglaw6180
    @meilinglaw6180 4 года назад +2

    I am from Asia , and I love you man , Your explanation is so great ‘ easy to understand

  • @dilinijayasinghe8134
    @dilinijayasinghe8134 6 месяцев назад +1

    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 !!!

  • @mateuszwojnarowicz7650
    @mateuszwojnarowicz7650 4 года назад +4

    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?

    • @xXxIMMORTALxXx
      @xXxIMMORTALxXx 4 года назад +2

      Check out his video titled: "Regression output Explained", he talks about ANOVA there.

  • @xiaomengchen3033
    @xiaomengchen3033 Год назад +1

    the girl at 2:00 is kim taeyeon right??? lolll

  • @aditigangar3133
    @aditigangar3133 3 года назад +2

    Thanks!

  • @deniz-gunay
    @deniz-gunay 2 года назад +2

    why didn't you divide (n-1) at 15:33 ?

    • @studyfrancais4989
      @studyfrancais4989 Месяц назад +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

  • @husinhhoa
    @husinhhoa 5 дней назад

    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.

  • @ARM26878
    @ARM26878 2 года назад +1

    Excellent video! Did u happen to cover proof/intuition behind why correlation lies between -1 and 1? Thanks

  • @Engineer78206
    @Engineer78206 3 месяца назад +1

    great stat teacher on yt

  • @QuietLumina
    @QuietLumina Год назад +1

    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!

  • @dineafkir5184
    @dineafkir5184 4 года назад +1

    Great videos. It would be nice with some time series :-)

  • @hassanrevel
    @hassanrevel 3 года назад +1

    B.T.w thanks a lot

  • @faheemali6720
    @faheemali6720 5 месяцев назад

    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?

  • @DrorProductiones
    @DrorProductiones Год назад +1

    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

  • @Ejun0
    @Ejun0 6 месяцев назад

    Could you please elaborate on why did you use the joint probability of X and Y when calculating the Variance of X?

  • @raphaelg.garces7409
    @raphaelg.garces7409 Год назад

    how do you get the (5.70) and the (1.87) in the CORR formula?

  • @MitsosDA
    @MitsosDA 4 года назад +1

    Hey Justin, Can you make some explanatory videos about panel data, fixed effects please? Thank you very much! Love your content!

  • @attilioturco
    @attilioturco Год назад

    did i miss the part where he explained why corr is cov/stdev? i mean what's the logic behind dividing by stdev?

  • @firstspar
    @firstspar 4 года назад +1

    Just wanted to say these are great videos and you are doing an amazing job producing these. Please keep them up!

  • @marinabaskakova2333
    @marinabaskakova2333 3 года назад +1

    So well put together and crystal clear! Thank you!

  • @vcello6450
    @vcello6450 2 года назад

    I should give YOU my tuition. Thank you sir!

  • @nia-nshaepcollins7468
    @nia-nshaepcollins7468 2 года назад

    Thank you so much for this explanation!!!!

  • @jacobblack2544
    @jacobblack2544 Год назад

    I bow down to your knowledge sir, Namastey!

  • @jamoliddinalimov7498
    @jamoliddinalimov7498 4 года назад +1

    Can someone explain how he got 5.70 and 1.87 at 8:25?
    Thanks in advance!

    • @KenSyRacing
      @KenSyRacing 4 года назад +1

      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

    • @ArthurC504
      @ArthurC504 3 года назад

      @@KenSyRacing when you calculate it , it's actually 5,15 and 1,67 tho

  • @virajsawant2619
    @virajsawant2619 2 года назад

    Can you please tell what is mean by interclass correlation coefficient

  • @obaid370
    @obaid370 Год назад

    can someone do SD of x and y in correlation, I am getting different figures?

  • @hassanrevel
    @hassanrevel 3 года назад

    Bro there's a correlation between stock market and temperature.

  • @sohamprajapati8784
    @sohamprajapati8784 2 года назад

    Amazing video! Thank you.

  • @jononolan946
    @jononolan946 8 месяцев назад

    Why cant you compare the numbers as is?

  • @Mr360kori
    @Mr360kori Год назад

    Sir what's segmental correlation

  • @tugastone
    @tugastone 4 года назад

    Excelent Videos, but I detected a problem here: -21/(7.06 * 3.146) is -0.9454877 and not -0.9448251. I think some confusion has been made between samples and populations

  • @jennymiao2623
    @jennymiao2623 Год назад

    Very good, thank you Justin.

  • @Puhazhenthi
    @Puhazhenthi Год назад

    Very useful, Thank you so much ❤️

  • @gameply347
    @gameply347 5 месяцев назад

    Is there an video for Z-scores or did i miss that?

  • @PraveenKumar-qn2zh
    @PraveenKumar-qn2zh Год назад

    Thank you so much clean explanation

  • @elijahkengura2021
    @elijahkengura2021 3 года назад

    Good lesson. I have earned 5mks already.

  • @kanewilliams1653
    @kanewilliams1653 3 года назад

    Isn't finding covariance of two variables, in that case, very similar to finding the regression of two variables? It's just finding how two variables are linked..

  • @elvinphilip3410
    @elvinphilip3410 2 года назад

    You are an awesome teacher!!!! Thanks.

  • @igor-yp1xv
    @igor-yp1xv 8 месяцев назад

    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!

  • @szilike_10
    @szilike_10 3 года назад

    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 : )

  • @jonykhan4395
    @jonykhan4395 Год назад

    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.

  • @krishnabarfiwala5766
    @krishnabarfiwala5766 4 года назад

    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

  • @marcussuzuki4031
    @marcussuzuki4031 4 года назад

    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.

  • @whetstoneguy6717
    @whetstoneguy6717 4 года назад

    Why are statisticians interested in covariance if it is not informative as correlation? EXCELLENT VIDEO.

  • @leeris19
    @leeris19 4 месяца назад

    THANK YOU FOR EVERYTHING

  • @ansonp.johnson4221
    @ansonp.johnson4221 2 года назад

    How did you calculate P(x,y)

  • @mvr192
    @mvr192 2 года назад

    Thank you, it was extremely informative 😁. You take your time and explain it very clearly.

  • @exploreformore3784
    @exploreformore3784 2 года назад

    Brilliant description on covariance

  • @morakano3355
    @morakano3355 Год назад

    Who needs university to study stat?

  • @mohamedelewa3694
    @mohamedelewa3694 3 года назад

    Why you do not divide the variance by n-1

  • @leonidasi-deas125
    @leonidasi-deas125 4 года назад

    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!!!!

  • @heatherpsyd
    @heatherpsyd 3 года назад

    Stats are not my friend. 🥺

  • @damieo8139
    @damieo8139 Год назад

    Thank you so so so much! I thank God for you. Keep doing what you are doing.

  • @myblankspacee
    @myblankspacee 2 года назад

    I love you man... You saved me 😭❤️

  • @shwetaadvani8941
    @shwetaadvani8941 3 года назад

    What a boon for people new to statistics .Thank you for explaining so logically !:)

  • @tanekakutitus50
    @tanekakutitus50 2 месяца назад

    perfect explanation👍

  • @tsholoMnice2019
    @tsholoMnice2019 7 месяцев назад

    Zedstatistics. You have out done yourself!!!👏👏👏

  • @taladiv3415
    @taladiv3415 2 года назад

    Your videos greatly aid understanding of a concept. Thank you for your contribution to society!

  • @I_THE_ME
    @I_THE_ME 3 года назад

    Thank you lots for the awesome video that actually focused on what can be accomplished by using the formulas and what they cosist of.

  • @pb3254
    @pb3254 7 месяцев назад

    Another great video explanation, reminds me of my A Level stats many moons ago!

  • @pdparmar
    @pdparmar 2 года назад

    Simply amazing!! Never thought it's so easy!! 🙏

  • @alibinshuja5390
    @alibinshuja5390 10 месяцев назад

    You are a gift to the world!

  • @ryany6478
    @ryany6478 3 года назад

    You explained better than my lecturer and textbook! Thank you.

  • @meilinglaw6180
    @meilinglaw6180 4 года назад +1

    A gifted teacher

  • @michelleraymundo7038
    @michelleraymundo7038 3 года назад

    Great explanation, I love the quick and dirty explanations.

  • @mrnkarun
    @mrnkarun 3 года назад

    How does P(X,Y) become equal to P(X)? and P(Y)=P(X,Y)

  • @akashvani628
    @akashvani628 4 года назад

    Hey. Can u make a video on step by step course for series of video on stats, to be consumed in your tutorial.

  • @miyamotomusashi4556
    @miyamotomusashi4556 2 года назад

    Thank you, just thank you.

  • @akhilchanganath4103
    @akhilchanganath4103 3 года назад

    Love from India ❤️

  • @arundhatisarkarbose5317
    @arundhatisarkarbose5317 4 года назад

    Thank you so much for the videos. They are truly helpful.
    Can you make videos for time series and panel data analysis?

  • @RiaziMohandesi
    @RiaziMohandesi 4 года назад

    Thank you very much Sir, one of the best explanation on the planet. From Iran

  • @rodlu811
    @rodlu811 Год назад

    Fantastic explanation!! Very intuitive and easy to understand .

  • @magtazeum4071
    @magtazeum4071 3 года назад

    Thank you very much!!!!!!

  • @BibosToys
    @BibosToys 3 года назад

    1:55 is that who i think it is lol

  • @ManikHussain-
    @ManikHussain- 2 года назад

    Just a boss public bro🥰

  • @gulcinmmmdova
    @gulcinmmmdova Год назад

    You're perfect teacher

  • @Competent00
    @Competent00 3 года назад

    Thank you so much. It was very helpful

  • @aminomidi7629
    @aminomidi7629 2 года назад

    Thank you for this useful content my friend.

  • @yoshifernando5264
    @yoshifernando5264 3 года назад

    This is such great prep before doing a business stats unit for the first time. Thanks heaps Justin!!

  • @kurmanzhanalimzhankyzy928
    @kurmanzhanalimzhankyzy928 2 года назад

    Thank you for all 🤍

  • @azboshippuden3770
    @azboshippuden3770 2 года назад

    Thank you for explaining this all easily!!

  • @James-md2xk
    @James-md2xk 3 года назад

    best stats teacher on youtube, thank you so much for your videos here