I have performed the statistical analysis and have not found how to perform the chart (the figure 1 of the papers in attached file). I need the commands (SAS or R) that permits me to perform the chart.
It gives me results in tables without the graph.
This resource has a good overview of how to chart the impact of a specific independent variable on your outcome variable within the contest of a multi-variable model.
The section on graphing single variables in a multi-variable model is on page 14, "Graphing a Multiple Regression".
https://www.pearsonhighered.com/assets/samplechapter/0/2/0/5/0205019676.pdf
This shows linear regression, so in the case of logistic regression you will expect the curves to have a sigmoid "S" shape when you plot them. Here is another resource which shows exact commands in R for plotting these data.
http://www.theanalysisfactor.com/r-glm-plotting/
For how long are you using R? I'm posting a very general approach to make a graph using ggplot2. There are other ways to do this but ggplot is a really nice package to construct graphs.
First create your model
model
Should be quite straightforward.
Key is to realize the chart is showing the predicted probability from the computed regression model vs a variable used in the model.
So you just get the predicted probabilities and plot them vs. their corresponding variable values.
# simulate dataset
set.seed(42)
n.obs
It seems like the ggeffects-package does what you're looking for:
https://cran.r-project.org/package=ggeffects
There are some examples here: https://cran.r-project.org/web/packages/ggeffects/vignettes/marginaleffects.html
which show how to use the package. The examples use a linear model, but you can use a lot of different models and the package will do the rest for you.
Example, which plots the predicted probabilities of y for predictor x1:
m
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis