Grouped boxplot with 2 y axes 2 variables per x tick

Question

I am trying to make a boxplot of cost (in Rupees unit) and installed capacity (in Megawatt unit) with xaxis as share of renewables (in % unit).

That is each x tick is associated with two boxplots, one is the cost and one of the installed capacity. I have 3 xtick values (20%, 40%, 60%).

I tried this answer but I get error that is attached on the bottom.

I need two boxplots per xtick.

from matplotlib import pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sns
plt.rcParams["font.family"] = "Times New Roman"
plt.style.use('seaborn-ticks')
plt.grid(color='w', linestyle='solid')
data1 = pd.read_csv('RES_cap.csv')
df=pd.DataFrame(data1, columns=['per','cap','cost'])

cost= df['cost']
cap=df['cap']
per_res=df['per']

fig, ax1 = plt.subplots()
xticklabels = 3

ax1.set_xlabel('Percentage of RES integration')
ax1.set_ylabel('Production Capacity (MW)')
res1 = ax1.boxplot(cost, widths=0.4,patch_artist=True)
for element in ['boxes', 'whiskers', 'fliers', 'means', 'medians', 'caps']:
    plt.setp(res1[element])

for patch in res1['boxes']:
    patch.set_facecolor('tab:blue')

ax2 = ax1.twinx()  # instantiate a second axes that shares the same x-axis
ax2.set_ylabel('Costs', color='tab:orange')
res2 = ax2.boxplot(cap, widths=0.4,patch_artist=True)
for element in ['boxes', 'whiskers', 'fliers', 'means', 'medians', 'caps']:
    plt.setp(res2[element], color='k')
for patch in res2['boxes']:
    patch.set_facecolor('tab:orange')

ax1.set_xticklabels(['20%','40%','60%'])
fig.tight_layout()  
plt.show()

sample data: data attached

Answer 1

By testing your code and comparing it to the answer by Thomas Kühn in the linked question, I see several things that stand out:

• the data you input for the x parameter has a 1-D shape instead of 2-D. You input one variable so you get one box instead of the three you actually want;
• the positions argument has not been defined, which causes the boxes of both boxplots to overlap;
• in the first for loop over res1, the color argument in plt.setp is missing;
• you have set x tick labels without first setting the x ticks (as cautioned here) which causes an error message.

I offer the following solution which is based more on this answer by ImportanceOfBeingErnest. It solves the issue of shaping the data correctly and it makes use of dictionaries to define many of the parameters that are shared by multiple objects in the plot. This makes it easier to adjust the format to your taste and also makes the code cleaner as it avoids the need for the for loops (over the boxplot elements and the res objects) and the repetition of arguments in functions that share the same parameters.

import numpy as np                 # v 1.19.2
import pandas as pd                # v 1.1.3
import matplotlib.pyplot as plt    # v 3.3.2

# Create a random dataset similar to the one in the image you shared
rng = np.random.default_rng(seed=123) # random number generator
data = dict(per = np.repeat([20, 40, 60], [60, 30, 10]),
            cap = rng.choice([70, 90, 220, 240, 320, 330, 340, 360, 410], size=100),
            cost = rng.integers(low=2050, high=2250, size=100))
df = pd.DataFrame(data)

# Pivot table according to the 'per' categories so that the cap and
# cost variables are grouped by them:
df_pivot = df.pivot(columns=['per'])

# Create a list of the cap and cost grouped variables to be plotted 
# in each (twinned) boxplot: note that the NaN values must be removed
# for the plotting function to work.
cap = [df_pivot['cap'][var].dropna() for var in df_pivot['cap']]
cost = [df_pivot['cost'][var].dropna() for var in df_pivot['cost']]

# Create figure and dictionary containing boxplot parameters that are
# common to both boxplots (according to my style preferences):
# note that I define the whis parameter so that values below the 5th
# percentile and above the 95th percentile are shown as outliers
nb_groups = df['per'].nunique()
fig, ax1 = plt.subplots(figsize=(9,6))
box_param = dict(whis=(5, 95), widths=0.2, patch_artist=True,
                 flierprops=dict(marker='.', markeredgecolor='black',
                 fillstyle=None), medianprops=dict(color='black'))

# Create boxplots for 'cap' variable: note the double asterisk used
# to unpack the dictionary of boxplot parameters
space = 0.15
ax1.boxplot(cap, positions=np.arange(nb_groups)-space,
            boxprops=dict(facecolor='tab:blue'), **box_param)

# Create boxplots for 'cost' variable on twin Axes
ax2 = ax1.twinx()
ax2.boxplot(cost, positions=np.arange(nb_groups)+space,
            boxprops=dict(facecolor='tab:orange'), **box_param)

# Format x ticks
labelsize = 12
ax1.set_xticks(np.arange(nb_groups))
ax1.set_xticklabels([f'{label}%' for label in df['per'].unique()])
ax1.tick_params(axis='x', labelsize=labelsize)

# Format y ticks
yticks_fmt = dict(axis='y', labelsize=labelsize)
ax1.tick_params(colors='tab:blue', **yticks_fmt)
ax2.tick_params(colors='tab:orange', **yticks_fmt)

# Format axes labels
label_fmt = dict(size=12, labelpad=15)
ax1.set_xlabel('Percentage of RES integration', **label_fmt)
ax1.set_ylabel('Production Capacity (MW)', color='tab:blue', **label_fmt)
ax2.set_ylabel('Costs (Rupees)', color='tab:orange', **label_fmt)

plt.show()

Matplotlib documentation: boxplot demo, boxplot function parameters, marker symbols for fliers, label text formatting parameters

Considering that it is quite an effort to set this up, if I were to do this for myself, I would go for side-by-side subplots instead of creating twinned Axes. This can be done quite easily in seaborn using the catplot function which takes care of a lot of the formatting automatically. Seeing as there are only three categories per variable, it is relatively easy to compare the boxplots side-by-side using a different color for each percentage category, as illustrated with this example based on the same data:

import seaborn as sns    # v 0.11.0

# Convert dataframe to long format with 'per' set aside as a grouping variable
df_melt = df.melt(id_vars='per')

# Create side-by-side boxplots of each variable: note that the boxes
# are colored by default
g = sns.catplot(kind='box', data=df_melt, x='per', y='value', col='variable',
                height=4, palette='Blues', sharey=False, saturation=1,
                width=0.3, fliersize=2, linewidth=1, whis=(5, 95))

g.fig.subplots_adjust(wspace=0.4)
g.set_titles(col_template='{col_name}', size=12, pad=20)

# Format Axes labels
label_fmt = dict(size=10, labelpad=10)
for ax in g.axes.flatten():
    ax.set_xlabel('Percentage of RES integration', **label_fmt)
g.axes.flatten()[0].set_ylabel('Production Capacity (MW)', **label_fmt)
g.axes.flatten()[1].set_ylabel('Costs (Rupees)', **label_fmt)

plt.show()