Logistic Regression Details Pt1: Coefficients


When you do logistic regression you have to make sense of the coefficients. These are based on the log(odds) and log(odds ratio), but, to be honest, the easiest way to make sense of these are through examples. In this StatQuest, I walk you though two Logistic Regression Examples, step-by-step, and show you exactly how the coefficients are derived and how to interpret them.

NOTE: This StatQuest assumes that you are already familiar with…
The main ideas of Logistic Regression:
Odds and Log(odds):
Odds Ratio and Log(odds ratio):
Linear Regression: and
Linear Models:

For a complete index of all the StatQuest videos, check out:

If you’d like to support StatQuest, please consider…
YouTube Membership:

…a cool StatQuest t-shirt or sweatshirt (USA/Europe):

…buying one or two of my songs (or go large and get a whole album!)

…or just donating to StatQuest!

Lastly, if you want to keep up with me as I research and create new StatQuests, follow me on twitter:

0:00 Awesome song and introduction
1:13 Review of Logistic Regression Concepts
2:47 Coefficients for continuous variables
10:46 Coefficients for discrete variables
17:52 Coefficients for combinations of variable types

#statquest #logistic

Nguồn: https://scoutdawson.com/

Xem thêm bài viết khác: https://scoutdawson.com/tong-hop/


  1. Song dedicated to nerds. In hawaiin beach watching statquest instead of getting pervy. Thats nerd alert(me included)

  2. Thanks Mr. Josh Starmer. I enjoy your videos. Do you have similar explanation on Multinomial Logistic Regression? Is there a video on Multinomial Logistic Regression? If not can you make one and add it to this series?

  3. Hello Sir, I really like your videos and the teaching style. You are really inspiring Sir. I also want to teach like you. I wanted to know how do you make your videos..I mean whuch software do you use to teach us?..I am totally naive but I also want to teach some concepts briefly which I know. Thanks

  4. Hello

    At 3:25…. you wrote as SIZE = 0.86+0.7*weight…

    What I know is y = mx+c

    I am unable to compare these 2 equations..you said Y = 0.86 , m = 0.7 that is OK. (How did you calculate m).

    But What about " x " and " c " …Please help.

  5. i am getting very confused at logistic regression. Why we need to use gaussian pdf in maximum likelihood estimator, why not use other pdfs like poisson or bernoulli. I am a beginner with ML. So pls anyone help me out here. I will be highly grateful

  6. want a live video of you singing. pls pls
    And you teach awesomest(cant find any word to appreciate, so made it up)

  7. josh which python package do you use to calculate the logistic regression Intercept and beta between a binomial dependent variable and a categorical variable. I used you calculation to estimate the beta and intercept of two variables, however i get different results if i use this python package

    model = smf.logit("left ~ C(department)", data = HR_Analytics).fit()

    Is this the correct package to use in python to calculate the coefficients for logistics regression model.

  8. The explanation is good and exciting foe me was that you calculating coefficient for continuous variable, but later on , i found explanation for calculating the categorical variables. can you tell me formula for calculating the continuous variable weight is this same as OLS here where log of odd is continuous

  9. 8:49 estimated, std err, z-val, p-val

    number of std dev the estimated intercept is away from 0 on a std normal curve
    if it's less than 2 std dev away from 0, it's not statistically significant

  10. Thanks for this nice demonstrative video. One question though, in terms of testing association between discrete variable and categorical outcome (example of gene mutation — obesity), can I consider that is exactly how Chi-squared test is performed on contingency table?

  11. OMG Josh, after watching your videos one by one that is Logs ,Odds and Log(Odds). Logistic Regression is just tip of iceberg Double BAMMM!. ( StatQuest – youtube infinity = StatQuest Infinity)

  12. Hi Josh thank you so much for the informative videos!! Statquest has always been my go-to resource for stats on the internet 🙂 Would you please correct me if this is the wrong interpretation of the geneMutant coefficient from around 16:40? Having a mutated gene is associated with a 2.35 increase in log of odds that the mouse is obese. Thank you!

  13. It's as if a base of 'e' is assumed for the log() function ..at least around the 6 minute mark? I was previously under the impression that unless explicitly noted as ln() — which of course uses 'e' — that a base of 10 applies to log()… Interesting.

  14. at 9:55 it talks about the estimated intercept: I don't get why this is estimated or what the standard normal curve that you go on to show represents (WHAT has a mean of zero?). don't we definitively know the intercept of a fixed line? or is it a "translation" of the standard normal curve for weights (or odds of weights?) into one of coefficients (seems nontrivial that it would remain a standard normal curve given the log aspect)?

  15. On 15:22 the left hand part of the equation should be log(odd-obese) rather than 'size'. Or i miss something?

    edit the time 🙂

  16. JOSH, just like your videos, your music is incredible. Thank you for all the efforts you put in. Quadruple BAM !!!


Please enter your comment!
Please enter your name here