Plotting density of logit and probit

tsrockon28 · November 9, 2017, 4:06pm

Hi,
I am trying to plot densities for the logit and probit models with plot function but it's not working out.
The code is below.
if ggplot2 can be used to achieve same outcome then it would be of great help.
dataset available here MP NSS.dta - Google Drive

MP_NSS$scst= MP_NSS$socgrp
MP_NSS$scst= ifelse(MP_NSS$socgrp== c(1, 2), 1, 0)
MP_NSS$pds2= MP_NSS$pds_wheat_qty
MP_NSS= MP_NSS %>%
  replace_na(list(pds2= 0))
MP_NSS$pds2= ifelse(MP_NSS$pds2>0,1, 0)
MP_NSS$total_land2= MP_NSS$total_land
MP_NSS= MP_NSS %>%
   replace_na(list(total_land2= 0))
logit3= glm(pds2~ scst+  unitvalue_rice_wheat+ total_land2+ I(total_land2* total_land2), family = binomial(link = "logit"), data = MP_NSS)
plot(MP_NSS$unitvalue_rice_wheat, MP_NSS$pds2)
curve(exp(coef(logit3)[1]+ (coef(logit3)[2]*(MP_NSS$unitvalue_rice_wheat)))/(1+ exp(coef(logit3)[1]+ (coef(logit3)[2]*(MP_NSS$unitvalue_rice_wheat)) )))

mara · November 9, 2017, 6:02pm

Hi @tsrockon28,

If you check out this post, Plotting your logistic regression model, it takes you through both the base and ggplot2 versions of making a chart from the model. The R Markdown document is also available:

github.com

rosemm/rexamples/blob/master/logistic_regression_plotting_part1.Rmd

# Plotting the results of your logistic regression Part 1: Continuous by categorical interaction
We'll run a nice, complicated logistic regresison and then make a plot that highlights a continuous by categorical interaction.

```{r generate_fake_data}
set.seed(24601) # setting this so the random results will be repeatable 

library(MASS)
covmat <- matrix(c(1.0,   0.2,   0.6, 
                   0.2,   1.0,  -0.5, 
                   0.6,  -0.5,   1.0), nrow=3) # the true cov matrix for my data
data <- mvrnorm(300, mu=c(0,0,0), Sigma=covmat) # generate random data that match that cov matrix
colnames(data) <- c("X1", "X2", "DV")
data <- as.data.frame(data)
data$group <- gl(n=3, k=ceiling(nrow(data)/3), labels=c("a", "b", "c", "d"))
# add some group differences and interaction stuff...
data$DV <- with(data, ifelse(group=="c" & X1 > 0, DV+rnorm(n=1, mean=1), 
                             ifelse(group=="b" & X1 > 0, DV+rnorm(n=1, mean=2) , DV)))
# make DV binary
data$DV <- ifelse(data$DV > 0, 1, 0)
head(data)

This file has been truncated. show original

tsrockon28 · November 9, 2017, 6:37pm

thanks.
but didn't help.
not getting typica llogistic curve.

mara · November 9, 2017, 6:42pm

If you can make this into a self-contained reprex, so others can see the code and output, that would be great. It's hard to tell what the problem is from description alone, and it's a tall-ask of others to download an external data file, etc.

Thanks

tsrockon28 · November 9, 2017, 7:39pm

http://localhost:17309/session/file77c22d7310c_reprex.html

tjmahr · November 9, 2017, 8:04pm

For curve() it wants an expression that uses the variable x. Replace MP_NSS$unitvalue_rice_wheat with x. Use add = TRUE to draw on top of the previous plot.

tsrockon28 · November 10, 2017, 5:00am

did that but getting weird curve

tsrockon28 · November 10, 2017, 8:47am

see this
http://localhost:17309/session/file77c65da6758_reprex.html

mara · November 10, 2017, 12:15pm

So, if you take a look at the URL here, you can see it begins with localhost, which means it's only on your computer (it's not a link to something other people can see on the web).

I definitely recommend taking the 10-15 minutes at some point to watch Jenny Bryan’s quick, helpful overview of the package and how to use it (Jenny starts ~10:40):

Others on these boards, including those new to R (looking at you, @christinelly ), can definitely attest to the worthiness (and relative ease) of getting reprex-ing down!

tsrockon28 · November 10, 2017, 12:21pm

Thanks Mara for all the help.
Margins package with cplot function achieves all this beautifully.