Legend for boxplot graph layers

Question

I've made this boxplot graph

# Tamanho do gráfico em polegadas
plt.figure(figsize=(11, 6))

# Gráfico
ax = sns.boxplot(x="CÓD. ESTAÇÃO", y ="FERRO", data=df_FERRO, linewidth=1.0, color='#5d6b75')

ax.set_ylim(min(df_FERRO['FERRO']) - 0.5, max(df_FERRO['FERRO']) + 0.5)

# Cores
colors = ['#8ECEEA', '#FDF7AD', '#AC5079']
bins = [0.3, 5, 20]
base = -10
for top, c in zip(bins, colors):
    ax.axhspan(base, top, color=c, zorder=0)
    base = top
# Adicionando Título ao gráfico
plt.title("Ferro dissolvido (2017 - 2022)", loc="center", fontsize=18)
plt.xlabel("Estação")
plt.ylabel("mg/L Fe")

plt.show()

I need to insert a legend for the layers (represented by colors), i may say it's a legend for the "colored boxes". I've got this example (graph made in R) of what I expect:

It doesn't need to be exact like this graph, but it needs to express this idea. I've only found examples for lines, instead of layers.

Answer 1

To get a legend in matplotlib, you typically add a label= keyword to the elements you want to pop up in the legend. Then, calling ax.legend() will add a legend with these elements together with the label text. The legend will appear in the same order as the elements are created in the plot. To have the highest label at the top in the legend, the code can be changed to draw that one first.

Here is an example:

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

# create some random test data
np.random.seed(777)
df_FERRO = pd.DataFrame({"CÓD. ESTAÇÃO": np.random.randint(1000, 1011, 100),
                         "FERRO": np.random.uniform(0.3, 5, 100)})

plt.figure(figsize=(11, 6))

ax = sns.boxplot(x="CÓD. ESTAÇÃO", y="FERRO", data=df_FERRO, linewidth=1.0, color='#5d6b75')
ax.set_ylim(min(df_FERRO['FERRO']) - 0.5, max(df_FERRO['FERRO']) + 0.5)

colors = ['#AC5079', '#FDF7AD', '#8ECEEA']
labels = ["high", "normal", "low"]
bins = [20, 5, 0.3, -10]
for base, top, c, label in zip(bins[1:], bins[:-1], colors, labels):
    ax.axhspan(base, top, color=c, zorder=0, label=label)

leg = ax.legend(title='legenda', framealpha=0.9)
for h in leg.legend_handles:
    h.set_edgecolor('black')

# set title and axes labels
ax.set_title("Ferro dissolvido (2017 - 2022)", loc="center", fontsize=18)
ax.set_xlabel("Estação")
ax.set_ylabel("mg/L Fe")

plt.show()