在 seaborn 散点图中对分类 x 轴进行排序

Question

I am trying to plot the top 30 percent values in a data frame using a seaborn scatter plot as shown below.

The reproducible code for the same plot:

import seaborn as sns

df = sns.load_dataset('iris')

#function to return top 30 percent values in a dataframe.
def extract_top(df):
    n = int(0.3*len(df))
    top = df.sort_values('sepal_length', ascending = False).head(n)

    return top

#storing the top values
top = extract_top(df)

#plotting
sns.scatterplot(data = top,
                x='species', y='sepal_length', 
                color = 'black',
                s = 100,
                marker = 'x',)

Here, I want sort the x-axis in order = ['virginica','setosa','versicolor'] . When I tried to use order as one of the parameter in sns.scatterplot() , it returned an error AttributeError: 'PathCollection' object has no property 'order' . What is the right way to do it?

Please note: In the dataframe, setosa is also a category in species , however, in the top 30% values non of its value is falling. Hence, that label is not shown in the example output from the reproducible code at the top. But I want even that label in the x-axis as well in the given order as shown below:

Answer 1

scatterplot() is not the correct tool for the job. Since you have a categorical axis you want to use stripplot() and not scatterplot() . See the difference between relational and categorical plots here https://seaborn.pydata.org/api.html

sns.stripplot(data = top,
              x='species', y='sepal_length', 
              order = ['virginica','setosa','versicolor'],
              color = 'black', jitter=False)

Answer 2

This means sns.scatterplot() does not take order as one of its args . For species setosa , you can use alpha to hide the scatter points while keep the ticks.

import seaborn as sns

df = sns.load_dataset('iris')

#function to return top 30 percent values in a dataframe.
def extract_top(df):
    n = int(0.3*len(df))
    top = df.sort_values('sepal_length', ascending = False).head(n)

    return top

#storing the top values
top = extract_top(df)
top.append(top.iloc[0,:])
top.iloc[-1,-1] = 'setosa'
order = ['virginica','setosa','versicolor']

#plotting
for species in order:
    alpha = 1 if species != 'setosa' else 0
    sns.scatterplot(x="species", y="sepal_length",
                    data=top[top['species']==species],
                    alpha=alpha,
                    marker='x',color='k')

the output is

Answer 3

For those wanting to make use of the extra arguments available in sns.scatterplot over sns.strpplot (size and style mappings for variables), it's possible to set the order of the x axis simply by sorting the dataframe before passing it to seaborn. The following will sort alphabetically.

df.sort_values(feature)