pandas plot histogram data frame index

Question

I have the following data frame (df) in pandas:

       NetPrice  Units  Royalty
Price                       
3.65    9.13    171    57.60
3.69    9.23     13     4.54
3.70    9.25    129    43.95
3.80    9.49    122    42.76
3.90    9.74    105    38.30
3.94    9.86    158    57.35
3.98    9.95     37    13.45
4.17   10.42     69    27.32
4.82   12.04    176    77.93
4.84   24.22    132    59.02
5.16   12.91    128    60.81
5.22   13.05    129    62.00

I am trying to create a histogram on the index ("Price) with an y-axis of "Units" . I started with the following:

plt.hist(df.index)

This gives me a histogram plotting the price. How can I add the Units to the y-axis? Right now it is just a "scale".

Thank you!

Answer 1

Because your data is already partially aggregated, you can't use the hist() methods directly. Like @snorthway said in the comments, you can do this with a bar chart. Only you need to put your data in buckets first. My favorite way to put data in buckets is with the pandas cut() method.

Let's set up some example data since you didn't provide some that's easy to use:

np.random.seed(1)
n = 1000
df = pd.DataFrame({'Price' : np.random.normal(5,2,size=n),
                   'Units' : np.random.randint(100, size=n)})

Let's put the prices into 10 evenly spaced buckets:

df['bucket'] = pd.cut(df.Price, 10)
print df.head()

      Price  Units           bucket
0  8.248691     98    (7.307, 8.71]
1  3.776487      8  (3.0999, 4.502]
2  3.943656     89  (3.0999, 4.502]
3  2.854063     27  (1.697, 3.0999]
4  6.730815     29   (5.905, 7.307]

So now we have a field that contains the bucket range. If you want to give those buckets other names, you can read about that in the excellent Pandas documentation. Now we can use the Pandas groupby() method and sum() to add up the units:

newdf = df[['bucket','Units']].groupby('bucket').sum()
print newdf
                  Units
bucket                 
(-1.122, 0.295]     492
(0.295, 1.697]     1663
(1.697, 3.0999]    5003
(3.0999, 4.502]   11084
(4.502, 5.905]    15144
(5.905, 7.307]    11053
(7.307, 8.71]      4424
(8.71, 10.112]     1008
(10.112, 11.515]     77
(11.515, 12.917]    122

That looks like a winner... now let's plot it:

 newdf.plot(kind='bar')