Logistic model solution in r

Question

I have fit a logistic model and I want to know the value of x when y=0.5.

library("faraway")

data = read.table("SBay.txt", header=T)
female = subset(data,sex==2)

amat = c(0,1)
mod1 = glm(amat[maturity]~age,family=binomial,data=female)
summary(mod1)
(len_50 = (log(1)-mod1$coef[1]) /mod1$coef[2])

(plot(female$age,amat[female$maturity],pch=1,col=2))
s = seq(0, 20, 0.1)
lines(s,ilogit(mod1$coef[1]+mod1$coef[2]*s),col=2,lty=2,lwd=2)

I google the answer and get:

F1 = ilogit(mod1$coef[1]+mod1$coef[2])
F2 = 0.5
solve(F1,F2)

I got the answer equal to 9.931613, however, it doesn't match the graph I've plot. SO I think there might be something wrong with my code. Here are my graph, it seem the answer is between 2 and 3.

https://i.stack.imgur.com/4qrAk.jpg

have you looked at `?predict.glm` – Andrew Haynes Nov 13 '17 at 13:53 — Andrew Haynes, Nov 13 '17 at 13:53

score 1 · Accepted Answer · answered Nov 13 '17 at 14:20

1

What you need is the predict function followed by an ifelse expression:

# Logistic Regression Classifier
mod1 = glm( amat[maturity]~age,family=binomial,data=female)

# Probabilities 
prob_pred = predict( mod1, type = "response", newdata = female) # probability outcomes

# Outcomes (True/False)
y_pred = ifelse( prob_pred > 0.5, 1, 0)  # NOTE: the 0.5, it filters outcomes > 0.5 probability

answered Nov 13 '17 at 14:20

hello_there_andy

2,039
2
21
51

some kudos here to @AndrewHaynes – hello_there_andy Nov 13 '17 at 14:21
1

How can I get the x value when y=0.5? – Z. Y. Jeng Nov 13 '17 at 15:54
Good question @Z.Y.Jeng, to do this you need to use the `predict()` function, with a numerical solver to find the X. Look at the first answer in this SO post: https://stackoverflow.com/a/32040556/3011648 – hello_there_andy Nov 15 '17 at 10:13
@Z.Y.Jeng, just some advice: It is often a good idea to share the background for the problem you like to solve. If you would edit this under a separate header it may lead to more up-votes. – hello_there_andy Nov 17 '17 at 15:23

Logistic model solution in r

1 Answers1