Variable selection procedures in R: Forward, backward, stepwise, and best-subsets regression (2020)

Поделиться
HTML-код
  • Опубликовано: 13 дек 2024

Комментарии • 60

  • @sandiprijal9521
    @sandiprijal9521 3 года назад +1

    You are the savior as all my classmates are using SAS and I am the only guy using R. Thanks Mike

  • @letho3110
    @letho3110 3 года назад +2

    Thank you so much Prof. Crowson for this very insightful video and also the additional educational materials. They have been very useful to me as a PhD candidate and I am very happy that I found your video.

    • @mikecrowson2462
      @mikecrowson2462  3 года назад

      You are very welcome! Best wishes in your studies!

  • @NEWGAMER-v5d
    @NEWGAMER-v5d 3 года назад

    Thank you very much. Not tried yet but looking for this since 2 days

  • @nilufarrashitova3254
    @nilufarrashitova3254 3 года назад +1

    Mike, you made a great job! Thank you so much!!! Explained in detail.

  • @MrAMerang
    @MrAMerang Год назад

    Thank you so much for sharing. I watched so many before I found your video and they were either using other softwares, not very clear or not covered the key points like the one at the end on "best subset" r function.

  • @jeffreylau2093
    @jeffreylau2093 4 года назад +3

    how is this process done for a logistic regression (binary) model?

  • @thebutcher573
    @thebutcher573 3 года назад

    honestly man thank you, you're a life saver

  • @shoumicshahid9315
    @shoumicshahid9315 4 года назад +1

    Hello, I have a question. Can we use this for ordinal logistic regression? I used mass package to perform my ordered logit model

  • @drryangagnon
    @drryangagnon 3 месяца назад

    @mikecrowson2462, I was reviewing some model selection stuff and once again you remain an amazing resource. Thank you!

  • @edmondmalepane3141
    @edmondmalepane3141 2 года назад

    dear mike , This line FWDfit.p

  • @manishadinesh2797
    @manishadinesh2797 Год назад

    Hi , can i get more information on the dataset what are we analyzing? can I get description of the variables?

  • @marjanhajizadeh5213
    @marjanhajizadeh5213 3 года назад +1

    Thanks you for the video .I have a huge data with 537 obs of 22 variables. I couldn't do the
    modcompare

  • @kelvinkipsang3600
    @kelvinkipsang3600 2 года назад

    Thank you so much,for deeper explanation

  • @galan8115
    @galan8115 4 года назад +1

    Good evening Mike, first of all thank you for the video :D. And second, does olsrr works with logistic regression? What i've read is just for linear models, (lm class, and not glm).

  • @hai-hoanguyen6653
    @hai-hoanguyen6653 4 года назад +1

    Great job! please me tell the package of "olsrr" works well with which version of R? 3.6.2 seems not to be suitable?

    • @galan8115
      @galan8115 4 года назад

      It works in mine, and i got R\R-3.6.2

  • @edmondmalepane3141
    @edmondmalepane3141 2 года назад

    Does this fit only the forward selection ?
    Using the o value method

  • @edmondmalepane3141
    @edmondmalepane3141 2 года назад

    Do you also fit lasso regression

  • @denSILVA25
    @denSILVA25 4 года назад

    Thank you very much! When I run FWDfit.p

  • @edmondmalepane3141
    @edmondmalepane3141 2 года назад

    what did i do wrong if um only getting the summary after fitting the forward selection

  • @muhammadsaleemkhan5761
    @muhammadsaleemkhan5761 2 года назад

    nice video, just wanted to know any similar package for Poisson regression analysis

  • @reshmawagh8355
    @reshmawagh8355 2 года назад

    Thanks a lot sir.... very informative..

  • @muhammadsaleemkhan5761
    @muhammadsaleemkhan5761 2 года назад

    nice video, is there any similar function for using Poisson regression

  • @sabrinaladouali5020
    @sabrinaladouali5020 2 года назад

    thank you so much sir , such a great video

  • @iPanxy
    @iPanxy 3 года назад

    Hello Mike, thank you for providing the best description of these methods! One question I have is...do I always just use the best subset method to find my model? If not, when should I use forward vs. backward?

  • @HarpreetKaur-bx1ej
    @HarpreetKaur-bx1ej 3 года назад

    When i am running ols step forward p i am getting na in two of rows which are in including parts of variable

  • @KareemLongwe
    @KareemLongwe 3 года назад

    Thanks for the video. How can I do variable selection for the linear mixed model

  • @odudeyusuf5644
    @odudeyusuf5644 3 года назад +1

    Pls The olsrr package doesn't work on my system. It's saying the package is developed under R 4.2.0.
    Any suggestions pls
    I've an assignment to submit by monday

    • @mikecrowson2462
      @mikecrowson2462  3 года назад +1

      I believe it should work. When you type library (olsrr) it will likely say what version of R it was developed on. This is generally the case when you call up any package. You should be able to use the olsrr functions after calling it up

    • @mikecrowson2462
      @mikecrowson2462  3 года назад +1

      Unless the package gets retired or degraded, it should work with later versions of R

    • @odudeyusuf5644
      @odudeyusuf5644 3 года назад +1

      Okay.
      I'll try and install the 4.2.0

    • @odudeyusuf5644
      @odudeyusuf5644 3 года назад +1

      Is it a must I get the same model when I used forward selection or backward selection

    • @mikecrowson2462
      @mikecrowson2462  3 года назад +1

      @@odudeyusuf5644 Forward and backward selection are two different empirically-based variable selection procedures. Sometimes they can yield the same model, but they can also yield different results. It really depends on your preference for selecting predictors for your model. You don't generally choose an approach based on what it gives you, but rather your reasons for a particular strategy for empirically-based entering or deletion of variables. Cheers!

  • @acpdevo
    @acpdevo 4 года назад

    Dr. Crowson Good work!

  • @bachlam9841
    @bachlam9841 3 года назад +1

    Hi Mike Crowson, your video provides the best explanation that I can find so far. Thank you a lot for making this video. I have some concerns related to applying this package to my project. Would you mind if I can contact you to ask for some advice?

  • @spencerantoniomarlen-starr3069
    @spencerantoniomarlen-starr3069 2 года назад

    I am getting completely different results, could this be because you didn't set a seed before running your regressions?

  • @DB-in2mr
    @DB-in2mr 4 года назад +2

    After this one Mike, (great again!) what about something on ridge, lasso elastic net regularization methods,... and ...then please make an entry on Machine Learning methods...courious to check out how you tackle them : ) Daniele

    • @mikecrowson2462
      @mikecrowson2462  4 года назад

      Hi Daniele, thanks for your message. Those are not procedures that I'm up on at this point. However, I DO enjoy learning new things, so I'll certainly see what I can do to oblige :)

  • @kevins6732
    @kevins6732 3 года назад

    How can you do this for logistic regression?

  • @ahmedelbeltagi6132
    @ahmedelbeltagi6132 3 года назад

    An attractive video show, thanks so much

  • @wasafisafi612
    @wasafisafi612 5 месяцев назад

    Thank you so much

  • @sarbajitg
    @sarbajitg 4 года назад

    Sir, what is the procedure for step-wise regression, does it starts from full model then add/remove or it starts from NULL model and then add/remove variables. In step-wise regression what is the sequence action I am little confused.

    • @yopomdpin6285
      @yopomdpin6285 3 года назад

      depend if you choose forward or backward selection

  • @wildvineyard
    @wildvineyard 4 года назад

    Very useful! Thank you!

  • @edmondmalepane3141
    @edmondmalepane3141 2 года назад

    Thank you

  • @ethanthealien
    @ethanthealien 4 года назад

    This is fantastic ! Thank you. =)

  • @shivc22
    @shivc22 2 года назад

    Isn't a smaller value of AIC better than a larger one?

    • @mikecrowson2462
      @mikecrowson2462  2 года назад

      Yes. When using Aic to compare models, the model with the smaller Aic is preferred.

  • @larryboateng1452
    @larryboateng1452 3 года назад

    Thank you for this

  • @krishnaiyer2556
    @krishnaiyer2556 3 года назад

    EXCELLENT

  • @refji8378
    @refji8378 4 года назад +1

    Extended regression models dear 😉🤝💪

  • @ishmanedu
    @ishmanedu 4 года назад

    good job done.

  • @drryangagnon
    @drryangagnon 2 года назад

    Hi Mike, a student was using this video for one of my classes and they shared the link for the regData is dead.

    • @mikecrowson2462
      @mikecrowson2462  2 года назад +1

      Hi Ryan, thanks for letting me know! It's driving me crazy that some of these links have stopped working. I've fixed the link under the video description, so your student should be good to go. I appreciate it!

    • @drryangagnon
      @drryangagnon 2 года назад

      @@mikecrowson2462 I definitely felt like a bit of a turd putting that comment in there just now. Thank you for fixing it though. I appreciate it very much!

    • @mikecrowson2462
      @mikecrowson2462  2 года назад

      @@drryangagnon Hi Ryan, it's all good. I appreciate you letting me know. You have a great evening!

  • @davemakafui4367
    @davemakafui4367 4 года назад

    thanks

  • @jolima2045
    @jolima2045 Год назад

    please how to do step with glmer? Is there a package?