CODEWITHSUNDEEP
pythonpandasmatplotlib
stackoverflowuser2010
stackoverflowuser2010

Reputation: 40869

Joining two Pandas dataframes and producing side-by-side barplot?

Suppose I have two Pandas dataframes, df1 and df2, each with two columns, hour and value. Some of the hours are missing in the two dataframes.

import pandas as pd
import matplotlib.pyplot as plt
data1 = [
    ('hour', [0, 1, 2, 4, 5, 6, 7, 8, 9, 10, 11, 12,
              13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23]),
    ('value', [12.044324085714285, 8.284134466666668, 9.663580800000002,
               18.64010145714286, 15.817029916666664, 13.242198508695651,
               10.157177889201877, 9.107153674476985, 10.01193336545455,
               16.03340384878049, 16.037368506666674, 16.036160044827593,
               15.061596637500001, 15.62831551764706, 16.146087032608694,
               16.696574719512192, 16.02603831463415, 17.07469460470588,
               14.69635686969697, 16.528905725581396, 12.910250661111112,
               13.875522341935481, 12.402971938461539])
    ]

df1 = pd.DataFrame.from_items(data1)
df1.head()
#    hour      value
# 0     0  12.044324
# 1     1   8.284134
# 2     2   9.663581
# 3     4  18.640101
# 4     5  15.817030

data2 = [
    ('hour', [0, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
              15, 16, 17, 18, 19, 20, 21, 22, 23]),
    ('value', [27.2011904, 31.145661266666668, 27.735570511111113,
               18.824297487999996, 17.861847334275623, 25.3033003254902,
               22.855934450000003, 31.160574200000003, 29.080220000000004,
               30.987719745454548, 26.431310216666663, 30.292641480000004,
               27.852885586666666, 30.682682472727276, 29.43023531764706,
               24.621718962500005, 33.92878745, 26.873105866666666,
               34.06412232, 32.696606333333335])
    ]

df2 = pd.DataFrame.from_items(data2)
df2.head()
#    hour      value
# 0     0  27.201190
# 1     5  31.145661
# 2     6  27.735571
# 3     7  18.824297
# 4     8  17.861847

I would like to join them together using the key of hour and then produce a side-by-side barplot of the data. The x-axis would be hour, and the y-axis would be value.

I can create a bar plot of one dataframe at a time.

_ = plt.bar(df1.hour.tolist(), df1.value.tolist())
_ = plt.xticks(df1.hour, rotation=0)
_ = plt.grid()
_ = plt.show()

enter image description here

_ = plt.bar(df2.hour.tolist(), df2.value.tolist())
_ = plt.xticks(df2.hour, rotation=0)
_ = plt.grid()
_ = plt.show()

enter image description here

However, what I want is to create a barchart of them side by side, like this:

enter image description here

Thank you for any help.

Upvotes: 2

Views: 2199

Answers (2)

ImportanceOfBeingErnest
ImportanceOfBeingErnest

Reputation: 339062

You can do it all in one line, if you wish. Making use of the pandas plotting wrapper and the fact that plotting a dataframe with several columns will group the plot. Given the definitions of df1 and df2 from the question, you can call

pd.merge(df1,df2, how='outer', on=['hour']).set_index("hour").plot.bar()
plt.show()

resulting in

enter image description here

Note that this leaves out the number 3 in this case as it is not part of any hour column in any of the two dataframes. To include it, use reset_index

pd.merge(df1,df2, how='outer', on=['hour']).set_index("hour").reindex(range(24)).plot.bar()

enter image description here

Upvotes: 2

araraonline
araraonline

Reputation: 1562

First reindex the dataframes and then create two barplots using the data. The positioning of the rectangles is given by (x - width/2, x + width/2, bottom, bottom + height).

import numpy as np

index = np.arange(0, 24)
bar_width = 0.3

df1 = df1.set_index('hour').reindex(index)
df2 = df2.set_index('hour').reindex(index)

plt.figure(figsize=(10, 5))
plt.bar(index - bar_width / 2, df1.value, bar_width, label='df1')
plt.bar(index + bar_width / 2, df2.value, bar_width, label='df2')
plt.xticks(index)
plt.legend()

plt.tight_layout()
plt.show()

plot

Upvotes: 1

Related Questions