79% of Regression Analysis Basics in under 18 Minutes [Simple, Multiple and Logistic Regression]

[@datatab]

If you like, please find our e-Book here: datatab.net/statistics-book

[@RendzoChabalala]

Thank you for making regression analysis so easy. ....connecting from south africa.

[@datatab]

Many thanks for your feedback! Regards Hannah

[@TravelerTriumphs]

So lovely listening to you. I have already shared your channel with my students. Just wish I could explain things the way you do. Thisd what I call it statistics made easy.

[@datatab]

Many thanks for your nice feedback and thanks for sharing!! Regards Hannah

[@abdelgaderalfallah]

Splendid, just SPLENDID ❤

[@datatab]

Many thanks : )

[@mohitgehlot6582]

Thanks for making this simple

[@marksegall9766]

At about 9 minutes, the word 'influence' should be replaced with 'predict'. Correlation data can not tell use if there is a causal relationship between two variables. It can be suggestive of an influence but we have to be careful about saying an independent variable causes a change in the dependent variable. We can say that the independent variable predicts the dependent variable.

[@datatab]

Hey, many many thanks for your feedback!!! That is true! I will pay attention to it in the following videos! Regards Hannah

[@narutofan7727]

how to use likerts scale will it be converted to dichotomous? before using logistic regression?

[@FRANKWHITE1996]

awesome video

[@datatab]

Many thanks : ) Regards Hannah

[@FRANKWHITE1996]

@@datatab no, no, no! It’s me who thank You! :)

[@radupopescu1985]

WRONG: what it is minimized is the vertical distance, the residuals, not the distance from the points to the line.

[@datatab]

Good point! In an introductory video, it’s definitely more common to focus on minimizing the vertical distances, or residuals, which is the most standard approach. However, there are indeed different ways to fit a regression line depending on the context and goals of the analysis. Thanks for your input-it's a great opportunity to clarify this for everyone! Regards Hannah

[@ShubhamMishra-tp1dv]

First viewer

[@datatab]

Greate : )

[@myworldfriends123]

😀🤩🤩🤩🤩🙂

[@datatab]

Thanks!

[@berdie_a]

Bad examples. I understand that you want to make this simple. However, please give better examples. The ice cream example is bad since observations should not be indexed by time. Otherwise, you run into the problem of autocorrelation, one of the assumptions you missed at the end. The money from cake example is bad since that situation is deterministic, not stochastic. You don’t need linear regression for that. You simply need a series of equations, to solve for them. Days of the weeks, temperature and sunshine are bad variables for multivar linear regression since all of those variables are correlated. Also, observations should not be indexed by time since you run into the problem of autocorrelation Lastly, wrong interpretation. When you are interpreting the coefficients you are speaking about the estimated average of the response, not the response itself. they are also interpreted given that the other variables are held constant.

[@datatab]

Hi many thanks for your feedback! But please have a look at the Video again. Regarading the Cake example we say in the Video: "But wait, that was too easy, we know how much a cake costs and how many fixed costs we have. What if we don't know a and b, like in the example with the number of ice cream sold and the temperature?" So we exactly say what you are criticitiong. So please, before you make such a harsh comment, make sure you watch the video you are criticising carefully! Regardsing the Ice Example: It is just to make a point, it is not to publish anything or to prove that a certain medicine works. So in theory there are perfect examples, but when you analyse data for the real world, the data is almost never perfect and yet you want to get as much insight as possible from the data. Therefore, it is your right to think that this is a bad example. But I would say that it's a matter of preference. SO, I would have preferred you to write: I personally don't find the example well chosen because... The way you wrote it, it sounds like one of the haters who just want to say something negative about it. Regards Hannah

[@berdie_a]

@@datatab I meant this as creative criticism for your next videos. Don’t take it personally. Always strive to be better. 1.) Even if you did not know b and a. The relationship between y and x is perfectly linear. In this case, you can simply take two data points to calculate the slope, and solve the intercept from there. Let me reiterate my point: there is no stochasticity in this process. The relationship between total money made and cake sales is perfectly deterministic, given that you make and sell the cake at constant prices. There is no need for linear regression in that example. 2.) On that point, the equation you gave on the video was the estimated model. A better illustration would have been to include the linear model itself, which includes the error term. This is to better explain why points deviated from the fitted line, and why “when you analyse data for the real world, the data is almost never perfect and yet you want to get as much insight as possible from the data.” 3.) As you’ve pointed out in the end, these models require assumptions to be met. You can’t just pick a bad set of variables (or data) and just run the model just because you want to milk insights despite knowing your estimates are going to be useless. Practice what you preach, do that in your examples. 4.) Not sure if it’s a preference but I can see you choosing more accurate examples. Strive to be better. Lastly, I see your passion for making these videos. However, it’s unfortunate to see that you’re being blinded by your emotions. Learn how to take criticism. You did not even acknowledge the wrong interpretation of the estimated coefficients and issues of autocorrelation and multicollinearity.

[@jasonandrewismail2029]

incorrect interpretation of regression

[@tonycincera3353]

This is a casual intro attempt to introduce regression, so please include your issues with this video before you complain. Oh, I’ve been a Data Scientist for 35 years and instructor at various levels of student knowledge of regression and Data Science and this isn’t a bad introduction for a novice. Include specifics for your comment, curious as to why you say it’s incorrect.

[@berdie_a]

⁠@@tonycincera3353 It does not hurt to be accurate and simple at the same time.

[@datatab]

Many thanks @tonycincera3353 , Regards Hannah

[@noyoo123]

[@datatab]

: )