9
$\begingroup$

I'm using the Kaggle Titanic dataset. One feature is "Embarked", the city the passenger embarked from. The survival rate appears to correlate with it, but I'm worried it may just be correlated with the ticket Fare (which the survival rate definitely correlates with).

I want to plot a histogram of the fares. That would be easy. However, I also want to, on the same plot, have the histograms for the three embarked values (Q,C,S), labeled by different colors.

I've searched but can't figure out how. I can achieve something relatively similar with FacetGrid:

g = sns.FacetGrid(train_df, col='Embarked')
g.map(plt.hist, 'Fare', bins=20)

which gives three separate histograms, almost what I want:

enter image description here

But that's not quite what I want. For example, it's not great for comparing them. I could calculate the averages, but ideally I'd see them overlaid, so the three histograms are on the same plot, and you can tell them apart because they're labeled different colors.

It seems like PairGrid is almost what I want, but it appears to be giving some strange axis for "Fare" (which is not between 0 and 1), and the legend is blank, and the colors are solid (so you can kind of only see the top layer).

g = sns.PairGrid(train_df[["Embarked","Fare"]], hue="Embarked",size=8)
g = g.map_diag(plt.hist)
g = g.add_legend()

enter image description here

I'm very confused and not sure how to even go about finding the right answer. It seems like Seaborn can do anything, some really cool things, but it's just impossible to find the specific thing I want to do for a given circumstance.

How could I achieve this?

Improve this question
$\endgroup$

1 Answer 1

Reset to default
4
$\begingroup$

With a pandas dataframe, the following should do it without seaborn:

train_df.groupby('Embarked')['Fare'].plot.hist(alpha=0.5)

If you have to use seaborn you can use FacetGrid without the col and row argument:

g = sns.FacetGrid(train_df, hue='Embarked')
g = g.map(plt.hist, 'Fare', alpha=0.5)
g = g.add_legend()
Improve this answer
$\endgroup$
4
  • $\begingroup$ Thanks, I might use that as a stopgap, but I'd really like to know how to do it in Seaborn :) $\endgroup$
    – GrundleMoof
    Commented Apr 23, 2018 at 22:16
  • $\begingroup$ @GrundleMoof like this? $\endgroup$
    – oW_
    Commented Apr 23, 2018 at 22:37
  • $\begingroup$ Thanks, but the thing I was asking about in my OP was for the histograms to be on the same plot (FacetGrid still creates 3 separate ones). $\endgroup$
    – GrundleMoof
    Commented Apr 24, 2018 at 0:37
  • 1
    $\begingroup$ @GrundleMoof have you tried the code above? it's one plot $\endgroup$
    – oW_
    Commented Apr 24, 2018 at 15:34

Your Answer

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

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