Seaborn: using boxplot cause running out of memory

Question

I would like to plot three boxplots for 1, 2 and 3 weight_cat values (these are the only distinct values it has). These boxplots should show dependency height on weight category (weight_cat).

So I have such a dataframe:

print data.head(5)

        Height    Weight  weight_cat
Index                                
1      65.78331  112.9925           1
2      71.51521  136.4873           2
3      69.39874  153.0269           3
4      68.21660  142.3354           2
5      67.78781  144.2971           2

The code below finally eats all my ram. This is not normal, I believe:

Seaborn.boxplot(x="Height", y="weight_cat", data=data)

What is wrong here? This is the link to manual. Shape of the dataframe is (25000,4). This the link to the csv file.

This is how you can get the same data:

data = pd.read_csv('weights_heights.csv', index_col='Index')
def weight_category(weight):
    newWeight = weight
    if newWeight < 120:
        return 1

    if newWeight >= 150:
        return 3

    else:
        return 2

data['weight_cat'] = data['Weight'].apply(weight_category)

Calling `Seaborn.boxplot(x = data.Height, y = data.weight_cat)` will cause same problem. I believe this is not the problem — Rocketq, Apr 16 '16 at 16:50
Are you calling some sort of show() function to bring it to the display? — Pax Vobiscum, Apr 16 '16 at 16:55
http://stackoverflow.com/questions/26597116/seaborn-plots-not-showing-up — Pax Vobiscum, Apr 16 '16 at 16:55
Are you importing matplotlib.pyplot to do this? What matplotlib version do you have? — Pax Vobiscum, Apr 16 '16 at 17:07

score 10 · Accepted Answer · edited Jun 20 '20 at 09:12

10

Swap the x and y column names:

import seaborn as sns
sns.boxplot(x="weight_cat" y="Height", data=data)

Currently, you are trying to create a chart with as many boxplots as there are different height values (which are 24503).

This worked for me with your data:

EDIT

If you want to display your boxplot horizontally, you can use the orient argument to provide the orientation:

sns.boxplot(x='Height', y='weight_cat', data=data, orient='h')

Notice that in this case, the x and y labels are swapped (as in your question).

edited Jun 20 '20 at 09:12

Community

1
1

answered Apr 16 '16 at 17:15

iulian

5,494
3
29
39

1

@Rocketq, I've updated my answer with a horizontal orientation example. Check it in case you would like a horizontal boxplot. – iulian Apr 16 '16 at 17:38
1

You could also give `weight_cat` a categorical datatype and then the orientation can be automatically inferred. – mwaskom Apr 16 '16 at 17:52
The `x`/`y` swap tip was crucial for me! I've spent half a day debugging my script... thank you! – jojeck Apr 22 '20 at 16:05

Seaborn: using boxplot cause running out of memory

1 Answers1

EDIT