As with Plotly, no Seaborn function enables us to directly plot a ridgeline. In order to do so, we inspired from this example displayed in Seaborn documentation, that makes use of a Seaborn FacetGrid object with kdeplots to generate a ridgeline graph. We adapted the data as well as some lines of code. Feel free to investigate by yourself how to customize this graph further!
# getting necessary libraries
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
sns.set_theme(style="white", rc={"axes.facecolor": (0, 0, 0, 0)})
# getting the data
temp = pd.read_csv('https://raw.githubusercontent.com/plotly/datasets/master/2016-weather-data-seattle.csv') # we retrieve the data from plotly's GitHub repository
temp['month'] = pd.to_datetime(temp['Date']).dt.month # we store the month in a separate column
# we define a dictionnary with months that we'll use later
month_dict = {1: 'january',
2: 'february',
3: 'march',
4: 'april',
5: 'may',
6: 'june',
7: 'july',
8: 'august',
9: 'september',
10: 'october',
11: 'november',
12: 'december'}
# we create a 'month' column
temp['month'] = temp['month'].map(month_dict)
# we generate a pd.Serie with the mean temperature for each month (used later for colors in the FacetGrid plot), and we create a new column in temp dataframe
month_mean_serie = temp.groupby('month')['Mean_TemperatureC'].mean()
temp['mean_month'] = temp['month'].map(month_mean_serie)
At this point, you can have a look at what the dataframe looks like
temp.head()
Date | Max_TemperatureC | Mean_TemperatureC | Min_TemperatureC | month | mean_month | |
---|---|---|---|---|---|---|
0 | 1/1/1948 | 10 | 8.0 | 7.0 | january | 4.493982 |
1 | 1/2/1948 | 6 | 4.0 | 3.0 | january | 4.493982 |
2 | 1/3/1948 | 7 | 4.0 | 2.0 | january | 4.493982 |
3 | 1/4/1948 | 7 | 4.0 | 2.0 | january | 4.493982 |
4 | 1/5/1948 | 7 | 3.0 | 0.0 | january | 4.493982 |
Eventually, we generate the ridgeline plot by first instantiating a Seaborn FacetGrid object.
# we generate a color palette with Seaborn.color_palette()
pal = sns.color_palette(palette='coolwarm', n_colors=12)
# in the sns.FacetGrid class, the 'hue' argument is the one that is the one that will be represented by colors with 'palette'
g = sns.FacetGrid(temp, row='month', hue='mean_month', aspect=15, height=0.75, palette=pal)
# then we add the densities kdeplots for each month
g.map(sns.kdeplot, 'Mean_TemperatureC',
bw_adjust=1, clip_on=False,
fill=True, alpha=1, linewidth=1.5)
# here we add a white line that represents the contour of each kdeplot
g.map(sns.kdeplot, 'Mean_TemperatureC',
bw_adjust=1, clip_on=False,
color="w", lw=2)
# here we add a horizontal line for each plot
g.map(plt.axhline, y=0,
lw=2, clip_on=False)
# we loop over the FacetGrid figure axes (g.axes.flat) and add the month as text with the right color
# notice how ax.lines[-1].get_color() enables you to access the last line's color in each matplotlib.Axes
for i, ax in enumerate(g.axes.flat):
ax.text(-15, 0.02, month_dict[i+1],
fontweight='bold', fontsize=15,
color=ax.lines[-1].get_color())
# we use matplotlib.Figure.subplots_adjust() function to get the subplots to overlap
g.fig.subplots_adjust(hspace=-0.3)
# eventually we remove axes titles, yticks and spines
g.set_titles("")
g.set(yticks=[])
g.despine(bottom=True, left=True)
plt.setp(ax.get_xticklabels(), fontsize=15, fontweight='bold')
plt.xlabel('Temperature in degree Celsius', fontweight='bold', fontsize=15)
g.fig.suptitle('Daily average temperature in Seattle per month',
ha='right',
fontsize=20,
fontweight=20)
plt.show()