1

I want to plot a bar and count plot in the same figure. But when I plot them with two different y-axis the two plot lay over each other (see picture). Is there an option to "shift" one of the plots to the side a bit?

fig, ax1 = plt.subplots(figsize=(15, 10))
ax2 = ax1.twinx()

ax1.bar(x=score[team], height=score[tot], color=colors)
ax2 = sns.countplot(x=dff[team], data=dff, palette=colors, order=team)

Data description

score[team] = ['a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm']

enter image description here

Community
  • 1
  • 1
Mirjam
  • 27
  • 9

1 Answers1

1

While a little messy but manually changing bar width and position can get the job done.
I have used seaborn's barplot it's easier to plot dataframes 

import seaborn as sns;sns.set()
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np

def change_width(ax, new_value,recenter=False) :
    for patch in ax.patches :
        current_width = patch.get_width()
        diff = current_width - new_value
        patch.set_width(new_value)
        if recenter==True:  
            patch.set_x(patch.get_x() + diff * .5) #To recenter the bars

df = sns.load_dataset('tips')
fig, ax1 = plt.subplots(figsize=(8, 8))
ax2 = ax1.twinx()

ax1 = sns.countplot(x='day', data=df)
change_width(ax1, 0.35)

ax2 = sns.barplot(x="day", y="total_bill", data=df,palette='viridis')
change_width(ax1, 0.35,True)

enter image description here The barplot's bar and countplot's bars can vary in height heavily, So I would say normalize the values, countplot doesn't support an Estimator so use barplot's estimator to find relative values 

estimator=lambda x: len(x) / len(df) * 100

The change_width function

Julkar9
  • 1,508
  • 1
  • 12
  • 24