CODEWITHSUNDEEP
pythonmatplotlibpandas
Mark Graph
Mark Graph

Reputation: 5121

Multiple plots on same figure with DataFrame.Plot

While I can get multiple lines on a chart and multiple bars on a chart - I cannot get a line and bar on the same chart using the same PeriodIndex.

Faux code follows ...

# play data
n = 100
x = pd.period_range('2001-01-01', periods=n, freq='M')
y1 = (Series(np.random.randn(n)).diff() + 5).tolist()
y2 = (Series(np.random.randn(n)).diff()).tolist()
df = pd.DataFrame({'bar':y2, 'line':y1}, index=x)

# let's plot
plt.figure()
ax = df['bar'].plot(kind='bar', label='bar')
df['line'].plot(kind='line', ax=ax, label='line')
plt.savefig('fred.png', dpi=200)
plt.close()

Any help will be greatly appreciated ...

Upvotes: 3

Views: 8172

Answers (1)

HYRY
HYRY

Reputation: 97261

The problem is: bar plots don't use index values as x axis, but use range(0, n). You can use twiny() to create a second axes that share yaxis with the bar axes, and draw line curve in this second axes.

The most difficult thing is how to align x-axis ticks. Here we define the align function, which will align ax2.get_xlim()[0] with x1 in ax1 and ax2.get_xlim()[1] with x2 in ax1:

def align_xaxis(ax2, ax1, x1, x2):
    "maps xlim of ax2 to x1 and x2 in ax1"
    (x1, _), (x2, _) = ax2.transData.inverted().transform(ax1.transData.transform([[x1, 0], [x2, 0]]))
    xs, xe = ax2.get_xlim()
    k, b = np.polyfit([x1, x2], [xs, xe], 1)
    ax2.set_xlim(xs*k+b, xe*k+b)

Here is the full code:

from matplotlib import pyplot as plt
import pandas as pd
from pandas import Series
import numpy as np
n = 50
x = pd.period_range('2001-01-01', periods=n, freq='M')
y1 = (Series(np.random.randn(n)) + 5).tolist()
y2 = (Series(np.random.randn(n))).tolist()
df = pd.DataFrame({'bar':y2, 'line':y1}, index=x)

# let's plot
plt.figure(figsize=(20, 4))
ax1 = df['bar'].plot(kind='bar', label='bar')
ax2 = ax1.twiny()
df['line'].plot(kind='line', label='line', ax=ax2)
ax2.grid(color="red", axis="x")

def align_xaxis(ax2, ax1, x1, x2):
    "maps xlim of ax2 to x1 and x2 in ax1"
    (x1, _), (x2, _) = ax2.transData.inverted().transform(ax1.transData.transform([[x1, 0], [x2, 0]]))
    xs, xe = ax2.get_xlim()
    k, b = np.polyfit([x1, x2], [xs, xe], 1)
    ax2.set_xlim(xs*k+b, xe*k+b)

align_xaxis(ax2, ax1, 0, n-1)

and the output:

enter image description here

Upvotes: 4

Related Questions