Box Plots in R with t-test, Mann-Whitney, ANOVA, Kruskal-Wallis, Shapiro-Wilk, Levene and post-hocs
HTML-код
- Опубликовано: 21 авг 2024
- Hi, this video builds on the previous one - Box-plots with statistical details, so I highly recommend to check it out first: • Box-Plots with Statist...
In this video you'll learn how to conduct the most important statistical tests and visualise them on top of informative box-violin-plots in a simple intuitive R syntax. Particularly, you’ll learn how to do Student’s and Welsh’ t-tests, Mann-Whitney, Fisher’s and Welsh ANOVA and Kruskal-Wallis tests … like the one you see on the screen… You’ll also learn how to choose a correct test via checking assumptions in R with Shapiro-Wilk normality test and Levene’s homogeneity of variances test. Last, but not least you’ll be able to get a correct post-hoc analysis for ANOVA and Kruskal Wallis with p-values corrected for multiple comparisons. If you don’t know what post-hoc and p-value corrections are, no worries, I’ll briefly explain it. So, this video is not only about simple R code, but also about learning statistics by doing. Let’s get right into it!
Don't hesitate to use english subtitles, I edited them.
If you only want the code (or want to support me), consider join the channel (join button below any of the videos), because I provide the code upon members requests.
I also created a quick demo in R in the next video.
Enjoy! 🥳
This comes from an outer level. THis is great. Thanks!!!
You're very welcome! I later actually produced a separate video for every test ggstatslot does. For example, you can check out t-test, or kruskal-wallis, or chi-square, or correlation ... any common statistical test.
@@yuzaR-Data-Science I am going to check them. Now I am really digesting the statistical information.
Nice presentation, thank you!
Thanks for a nice feedback, Vladimir! I am glad it is useful!
Awesome stuff. As always.
I want a vid from you on ElasticNet and how it can supplant Ridge/Lasso.
Great suggestion! The machine learning stuff is coming anyway. But it'll take some time. I just need to cover some basic .... and get more time from my day job. Year, similar to the tests, I wanna cover every useful model. Ideally :)
@@yuzaR-Data-Science MARS is interesting too. Does feature selection/reduction and is kinda interpretable.
Oh, that's another good one I have to dive into to be able to answer. We'll get there ;)
thanks you very much to teach
Thanks for watching
excelente video gracias
de nada amigo
Excelent.
Thanks.
Glad you like it!
good work man, any chance of advising me? got some data...Kinda want more than Wilcoxon from SPSS
Thanks for the feedback, mate! Check out my later videos and blog-post (links to blogs are always below the video in the description). You’ll get many ideas. Unfortunately I can’t advise folks, have two day jobs and making videos on top of it. But I totally recommend you to switch from SPSS to R
@@yuzaR-Data-Science I just need to know how do I swap places of two variables (boxes) on the plot. After on the left, makes no sense... Thanks.
@@yuzaR-Data-Science Also, where do we find SD
No worries, mate, you just need to change the levels of a factor variable "test" in your data:
Fitness %>%
mutate(test = factor(test, levels = c("Before", "After")))
the test will use the order you define in your data
you mean standard deviation of every group? well, there is an amazing function "describe" from {dlookr} package. It gives you all the stats for your groups in one word. it works perfectly with "tidyverse", thats how you can use "group_by()". I also did a video on dlookr, there are more useful things then just "describe"
Sadly, but notch is not accessible now
Yeah, I asked the author of the package about it, and here is what he said: "I removed notch because it often just collapses on itself (mathworks.com/matlabcentral/answers/461791-what-do-lines-that-double-back-on-themselves-mean-box-plots)- Additionally, it is typically used for significance testing, which is redundant because ggstatsplot already provides such details in the plot itself."
@@yuzaR-Data-Science Thank you for your answer.
You are welcome!