CODEWITHSUNDEEP
pythonmatplotlib
M_Shimal
M_Shimal

Reputation: 433

Custom sorting of categorical variable on x-axis using matplotlib

I would like to order the months on x-axis to the order I specify. I have googled extensively, to learn how to do this but to no luck. I am very familiar with R language, and I would do this very easily in R using factor class and it's levels. But I am relatively new to python and what I have learned from reading is that Categorical dtype in python is the closest to factor in R. However, there seems to be a major behavioral difference to these classes in two language. There is no sorting of categorical order when plotted using pyplot.bar() but the same plot is ordered correctly in seaborn bar plot.

Is there an option for custom ordering of categorical variable in a dataframe for pyplot.bar()?

pandas = 0.22.0
matplotlib = 2.1.2
seaborn = 0.8.1



import pandas as pd
import matplotlib.pyplot as plt
from pandas.api.types import CategoricalDtype


TestData = pd.DataFrame({'value':[1,2,5,3,5,6,8,9,8,1,2,8,9],'Month':['Jan','Mar','Jan','Feb','May','Apr','Jan','Mar','Jan','Feb','May','Apr','May']})

# Applying custom categorical order
MonthLabels = ['Jan','Feb','Mar','Apr','May']
M_catType = CategoricalDtype(categories = MonthLabels, ordered = True)
TestData['Month'] = TestData['Month'].astype(M_catType)

plt.bar('Month','value', data=TestData)

enter image description here enter image description here

SOLVED

May have been an error with the version of matplotlib. I updated the version to 2.2.2 after reading this post and everything worked as expected(i.e, axis is sorted to the order provided when setting categories. Also I set the category using the code below, 

TestData['Month'] = pd.Categorical(TestData['Month'], categories = MonthLabels , ordered = True)

Upvotes: 11

Views: 12117

Answers (3)

6magics
6magics

Reputation: 11

This is what worked 4 me:

MonthLabels = ['Feb', 'Mar', 'May', 'June', 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec']
palette = ['#ff9100', '#ad0000']

# Use pd.Categorical to order the columns

df1 = df.copy()
df1['Month'] = pd.Categorical(df1['Month'], categories=MonthLabels)

sns.histplot(data=df1, x='Month', palette=sns.color_palette(palette, 2), hue='Revenue', edgecolor='#FFF', kde=True, legend=True)
plt.suptitle('Distribution of Revenue by Month')
plt.show()

Upvotes: 1

BioNinja
BioNinja

Reputation: 81

The only way works for me is to set xunits to the desired order

import pandas as pd
import matplotlib.pyplot as plt
from matplotlib.category import UnitData
TestData = pd.DataFrame({'value':[1,2,5,3,5,6,8,9,8,1,2,8,9],
                         'Month':['Jan','Mar','Jan','Feb','May','Apr','Jan','Mar','Jan','Feb','May','Apr','May']})
fig, (ax1, ax2) = plt.subplots(1,2, figsize=(8, 4))
# Applying custom categorical order
MonthLabels = ['Jan','Feb','Mar','Apr','May']
bar1 = ax1.bar('Month','value', data=TestData)
# set xunits with UnitData
bar2 = ax2.bar('Month','value', data=TestData, xunits=UnitData(MonthLabels))

enter image description here

Upvotes: 3

eduffy
eduffy

Reputation: 40224

This might help; from the documentation:

Note New categorical data are not automatically ordered. You must explicitly pass ordered=True to indicate an ordered Categorical.

Upvotes: 2

Related Questions