Seaborn Hue and Size Legend

Question

I'm trying to change the Bins in the legend of a plot created with Seaborn. The data is from 0 to 100, however Seaborn bins from 1 to 120 and 0 to 80. I've tried using hue_norm and size_norm but to no avail.

Please see code below and picture attached:

import matplotlib.pyplot as plt
import seaborn as sns
from import_portfolio import df, portfolio
import numpy as np
from adjustText import adjust_text

df_factor = df[df.columns[df.columns.str.contains('Factor Percentile')]]

columns = []
for column in df_factor.columns:
    split = str.split(column, sep=' ')
    fac = split[split.index('Factor') - 1]
    columns.append(fac)

df_factor.columns = columns
df_factor['Symbol'] = df['Symbol'].copy()
df_factor = df_factor.replace(' ', np.NaN)

plt.figure(figsize=(13,7))
ax = sns.scatterplot(data=df_factor.dropna(), x='Valuation', y='Quality', hue='Momentum', size='Growth', palette='RdYlGn', sizes=(20,150))
plt.xlim(0,100)
plt.ylim(0,100)

adjust_text(texts, arrowprops=dict(arrowstyle='-', color='k', lw=0.5))

plt.legend(bbox_to_anchor=(1.05,1), loc=2, borderaxespad=0.)

Picture:

Anyone any idea how to resolve this issue?

Thank you

Answer 1

You can alter the legend, but since you are working with a data.frame and seaborn, one option is to make your hue and size categorial to start, with provide the matching label or colors to sns.scatterplot

For example:

import matplotlib.pyplot as plt import seaborn as sns import numpy as np

np.random.seed(999) df_factor = pd.DataFrame(np.random.uniform(0,100,(20,4)),columns=['Valuation','Quality','Growth','Momentum'])

Here we introduce another column that discretizes the two columns. You can also overwrite it or use a function. Below I use pd.cut to assign values between 0 to 20 (including 20) to have a label 20, 21-40 to have a label40 and so on:

df_factor['Growth_lvl'] = pd.cut(df_factor['Growth'],[0,20,40,60,80,100],labels=[20,40,60,80,100])
df_factor['Momentum_lvl'] = pd.cut(df_factor['Momentum'],[0,20,40,60,80,100],labels=[20,40,60,80,100])

plt.figure(figsize=(13,7))
ax = sns.scatterplot(data=df_factor.dropna(), x='Valuation', y='Quality', hue='Momentum_lvl', 
                     size ='Growth_lvl', palette='RdYlGn',
                     sizes = list(np.arange(10,100,20)),
                     hue_order= [20,40,60,80,100])
plt.xlim(0,100)
plt.ylim(0,100)

plt.legend(bbox_to_anchor=(1.05,1), loc=2, borderaxespad=0.)