CODEWITHSUNDEEP
rlogistic-regression
superbot
superbot

Reputation: 461

Is it possible to plot logistic regression with categorical variables as independent variables?

I have tried both r plot and ggplot. They don't allow plotting logistic regression curve when you have categorical variables as independent variables (x-axis). When I tried after converting the categorical variables to random numbers, it worked. But that's confusing. Is there any solution, or am I missing something? Thank you in advance.

For example:

g <- ggplot(decision_use, aes(x=decision, y=use)) + geom_point(alpha=.1) +
  geom_smooth(method = "glm", 
    method.args = list(family = "binomial"), 
    se = FALSE)

and

plot(decision, use)
g=glm(use~decision,family=binomial, decision_use)
curve(predict(g,data.frame(decision=x),type="resp"),add=TRUE)

With decision as types of people and use as 1 or 0.

Upvotes: 2

Views: 6248

Answers (2)

Mattia Da Campo
Mattia Da Campo

Reputation: 41

I've been using this package that gives you great effects plots.

Let LogitModel be your Logistic Regression model 

install.packages("effects") # only need to do once. 
library(effects)
plot(allEffects(LogitModel))

Hope this helps

Upvotes: 3

Lyndon Walker
Lyndon Walker

Reputation: 105

Here is a great set of examples https://data.library.virginia.edu/visualizing-the-effects-of-logistic-regression/ It doesn't use ggplot but has an example of effect of a categorical variable among the examples.

One with ggplot https://blogs.uoregon.edu/rclub/2016/04/05/plotting-your-logistic-regression-models/

Upvotes: 1

Related Questions