1
$\begingroup$

I have a dataframe in the following form:

|Regions | F1 | F2 | F3 | ... | Fn |
|Region_1| X11| X12| X13| ... | X1n|
|Region_2| X21| X22| X23| ... | X2n|
...
|Region_k| Xk1| Xk2| Xk3| ... | Xkn|

Where every row represents an Italian region and every column is a feature that the individuals living in that region have. The value $X_{ij}$ is the number of individuals living in the region i and having the feature j, so the sum of each row gives the popolation living in that region; the sum of each column gives the people having that feature.

Boxplot depicts groups of numerical data through their quartiles. It is correct to create a boxplot on the columns of my data? It gives the correct information or I have to apply some transformation on my data (e.g., scaling, normalization, etc)?

$\endgroup$
  • $\begingroup$ Every Xij is an absolute frequency. $\endgroup$ – darioSka Nov 26 '16 at 11:45
  • $\begingroup$ Do you want to summarise each column or do you want to compare the distributions? $\endgroup$ – mdewey Nov 27 '16 at 13:44
  • $\begingroup$ I want to summarise each column $\endgroup$ – darioSka Nov 27 '16 at 14:50
2
$\begingroup$

The Boxplot will aggregate your values by columns so if you want to know for each feature the distribution of people with that characteristic, you can do it and it could be useful, but it's still not clear to me what do you mean with "it make sense and it gives the correct informations".

$\endgroup$
  • $\begingroup$ these data are absolute frequencies, it is correct to use a boxplot on them? $\endgroup$ – darioSka Nov 26 '16 at 11:42
2
$\begingroup$

To Fracesco's answer I would add that if features are measured in different scales (example: height, weight, income, number of children...), each boxplot will make sense in its own but comparing them in the same graph and the same scale won't make sense - and that is what boxplot(myDataFrame) does.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.