What to plot in order to pre-analyze the data before building a logistic model?

Ask Question

Asked 7 years, 10 months ago

Modified 5 years, 11 months ago

Viewed 124 times

In linear models I was able to do a scatter plot of just the response values and see, whether they're distributed "reasonably".

However, in logistic models the response is a binary variable and thus produces a kind of difficult to interpret plot of just values $0$ and $1$.

What kind of plots can I do in order to pre-analyze the data before doing a logistic model on it?

kjetil b halvorsen♦

81.7k32 gold badges199 silver badges654 bronze badges

asked Nov 25, 2016 at 19:18

mavavilj

4,1195 gold badges28 silver badges56 bronze badges

$\begingroup$ Do you absolutely need a plot? Why not do cross-tabulations, comparison of means, etc.? You can also do histograms or box-plots over the two categories of the outcome variable. Depends on the purpose of doing the plots/bivariate analysis really $\endgroup$
– Marquis de Carabas
Commented Nov 25, 2016 at 21:09
$\begingroup$ @MarquisdeCarabas No those would be fine as well. I just used plots as an example. $\endgroup$
– mavavilj
Commented Nov 26, 2016 at 10:20
$\begingroup$ Examples, with explanation, were provided at stats.stackexchange.com/questions/138605, stats.stackexchange.com/questions/37683, stats.stackexchange.com/questions/14480, and stats.stackexchange.com/questions/69820. Many of these will work even for exploratory data analysis ("pre-analysis"). $\endgroup$
– whuber ♦
Commented Oct 8, 2018 at 15:05

Add a comment |

0 Your Answer

Name

Required, but never shown

Post as a guest

Name

Required, but never shown

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.