I have said it before and I will say it as many times as it needs to be, Josh, you are a blessing. Thank You. Usually there are certain small details which our professors often skip and those details are the ones upon which advance systems are based but you lacked the initial concept thus it gets difficult. But you put and explain every detail beautifully.
I appreciate the explanation of the binomial distribution without using the example of flipping a coin (1000 times!!), instead you used a real-world example, it helped me understand the basic concepts of the binomial distribution. You did an amazing job.
The way you have broken down the nasty looking formula into easily understandable and relatable stuff is mind blowing . Thanks for creating such videos. Super helpful tht waz. Pls do more of this sort :)
It blows my mind thinking that future human generations will have such a great knowledge source on RU-vid, I think generations to come will learn statistics from here, you will be remembered as a legend, sir!!
"The binomial distribution will tell us what to expect if there is no preference"- that statement just tied up in one string all the scattered information I had so far about binomial distributions. Thank you for this elegant explanation.
These videos blow my mind everyday . Again its 5 am and i am studying stats quest. I'll summarize my understanding for others to see here . What you learned in p value videos about the two sided p-value and the video on 'what are stats model' was used here . Here our model(which was also our null hypothesis) is that both flavours are EQUALLY loved , so both have a probability of 0.5(this is NOT p-value, this is just our model-Remember that a model can be an 'equation' or a 'relationship' between variables) And since are p-value came out to be larger than 0.05 we rejected the alternate hypothesis that one of them is more loved. Our p value was 1 so we FAILED to reject the null hypothesis. BAM!
Hi, Josh, absolutely fantastic work! I was wondering if you could do something on Bayesian statistics? First off a frequentist vs bayesian perspective video, then a finding likelihood, prior, and posterior series!?
Hye Mandeep, I love this video too..... Btw, I have just posted about how to find probability using Binomial Distribution.....i hope it's helpful...... 💓
Great video. At the beginning, I was a bit scared by the equation of bionomial distribution, but you have explained it very clearly and in the end it resulted to be very easy. Thank you very much for this video.
Thank you for these Josh. You are doing an amazing job of making this stuff more accessible to people like me, struggling to improve and grow in my career :)
Love your vids because they don’t use much terminology and explain it in the simplest way 🙌 Using diagrams to explain instead of spouting paragraphs makes stats so much easier to understand!
Hi Josh, can you please please clarify the following: Why we don't we use combination instead of permutation...although I know the value would be same for both in this case where x = 1.......asking this because here ordering is not important so why (N)P(x) instead of (N)C(x)....I would be satisfied if I get to know this
Permutations keeps track of all the different ways that 3 people can like grape fanta and 4 people can like orange fanta. Combinations does not do this.
i am a bit confused. when we want to know if nearly half the population eg (4/7 or 3/7 in this example) like orange, wont we get p-value of 1 every time, irrespective of prob of orange? P(O) can be 0.7 instead of 0.5 and we will get the same p-value. because then the double -sided p-value will include all possible outcomes
Remember, the double sided p-value only includes equal or rarer events. So if the hypothesis is that the probability of something happening is 0, and it happens, then we will reject the hypothesis because it says that the probability of it happening is 0, and an equally rarer event would also have probability 0, and all rarer events would also have probability 0, so the p-value = 0.
This has to be the first comment (in a long time) that I've been inspired to write because of the sheer number of "light bulb" moments your tutorials give me. Great going and thank you!
I don’t if my opinion is inversely correlated to yours. But statquest is not cool….. It is the coolest in the entire statistics universe found on RU-vid.
Bro you don’t understand how much this video helped me 😭 I’m a homeschooler who’s in 9th grade and this was the work my mom gave me for today. Since I plan on taking college classes very soon, she wants to get me started on harder math problems. I was literally on the verge of tears while trying to use the nasty looking way for this. Thank you
This video helped me understanding a lot of things, nevertheless, I got confused with the "p-value = 1", I thought that empirical p-values can never be 0 or 1, only the values in between; also, I don't get how the provability of an uneven event can be the entire probability space. I'm just slow for stats.
first of all, StatQuest is fantastic! One question: how do I compare if two binomial distributions are different from each other? What test should be applied?
Oh my man, Oh my man, i just dont know how to thank you, i just dont know how to thank you....let me be a living proof of it....the videos you have made are so well corelated that the moment you explain something new in new video my brain automatically knows, goes and wanders to all those past videos where it need to go to have a recall....its so good....thank u thank u thank u
Hi. Thank you for the great video! I'm a bit stuck, would be great if you could give me some guidance! :) 11:20 the p-value of 4 out of 7 people preferring orange fanta is 0.273 14:07 the p-value of 4 out of 7 people preferring orange fanta is 1 It seems the subjects are the same but the numbers computed are different... why do we need the one at 14:07? Thanks!!
p-values are different from probabilities (and I know that is confusing because it seems like "p" should stand for "probability"). However, they are different. To learn more about p-values, see: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-vemZtEM63GY.html and ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-JQc3yx0-Q9E.html
That moment where you randomly looking for someone to explain you binomial distribution and out of blue you click on some sodas videos and it turns out to be StatQuest video so now you can be sure you will'be explained the material well so you can invest saved time into procrastinating while buying some third party drink that doesn't fit into the model unless we deny any Fanta but the true one. BAM!
In binomial Distribution "p" (in video p is someons will take orange Fanta) it will be always 0.5 as we are taking only 2 value so why ? We denote it as p why we directly not write 0.5 ?
Although 'p' is frequently set to 0.5, it isn't always set to 0.5. For example, in 2004, the ratio of boys to girls in China was 121 boys for every 100 girls. Thus, if we were to look at gender in China, we would set p=121/(121 + 100) = 0.55 (if 'p' is for boys, if it's for girls, p=0.45).
Love the video Josh. Just the words Binomial Distribution used to scare me. You have explained so well. Kudos! But, just one thing. I've never met a person who actually loves Grape Fanta. Only in the world of Stats will someone equally prefer Grape Fanta to Orange Fanta.
Thank you! Would you please help me clarify the confusion? Is binomial distribution just let us count the chance of each situation? I'm not sure why or when I use this for test. With the demonstrate, I only know no matter how much sample amounts, only variety a little not mean preference. But can this test for: If I get 1 orange, 6 grape, have any preference? For after count I get pr= 0.055 p-value=0.126 I don't know how to use the result for it >0.05 but I feel it have preference. Or I should think the "possibility" original is 0.5, so If p-value lower than "possibility" than there's preference?
The test becomes more useful with a larger sample size. If we had 1 orange and 10 grape, then the test would confirm your intuition that there is a preference.
probability that 4 out of 7 people prefer orange fanta over grape fanta 0.273 under the assumption that probability of fanta or grape is 0.5. and the ending of the pvalue just proves that. equally preferred, because we assumed that probability of chosing orange fanta is 0.5 and probability of chosing sprite is 0.5. Without pvalue, we can say that both are equally preferred.!!
What we're doing is testing the assumption that they are equally preferred and the data (4 of 7) does not convince us to reject that assumption (because the p-value is so large). Alternatively, we could test the assumption that only 0.0001% of people prefer orange. In that case, the data may convince us to reject that assumption because the p-value might be very small.
A probability distribution can also be used to calculate likelihoods - which sounds like I'm saying the same thing...but in statistics, likelihoods and probabilities are different and we do different things for them. Likelihoods are often used to fit a distribution to a collection of measurements.
Hi Josh, great video as usual! Moreover, I love your songs titled 'Saturday' and 'A Drink from the Well'. Didn't know you were gifted musically as well :) I'll be buying that album soon, and I encourage others to have a listen too! What's your favourite song, Josh?
Thank you very much!!!! I'm really glad you like my songs. I used to post a new one every month, but people complained, so I stopped. However, I still record a new song every month. I've been doing this for 98 months so far! :)
I have a question : in the correlation video ( you said that a p-value tells us the probability that a randomaly drown dots will result in a similary strong relashinchip or stronger) so that means if my p-value is small then the I have more confidence in the line that fits the data ...... and in this video when the p-value was 1 we said that the bionomial distribution with p = 1/2 is a good fit for the data .... so how is that? Am I missing something?? Thank you in advace
With correlation, the p-value compares how well a line fits the data to a random line. When the p-value is small, the probability that a random line would fit the data as well, or better, is very small, so we reject the "null" (that a random line would be a good model) and that tell us that our specific model (the line we fit to the data) is a good model. In contrast, in this case (with orange vs grape fanta), our model is that it's random whether or not someone will like orange or grape. In otherwise, we could just flip a coin and if it lands heads, then we could make a relatively good guess that a person will like orange fanta. So, the large p-value fails to reject the "null", that it is just a random choice. So our model is good.
I wanted to ask, I am little bit confused. I remember our professor told that order matters, even if they are same. So why we don’t consider OrangeF1 and OrangeF2 with different order? But the sun then 1, so it doesn’t make sense… ahhh, I am so confused ;c
