Spread the love

Learn how to make beautiful GUI apps

Comprehensive, for Kivy beginners, easy to follow.

Get the book here (PDF) or on Amazon:

ebook / paperback (black and white) / paperback (full color)

IMPLICIT CREATION OF MULTI-INDEXED OBJECTS

In the previous couple articles in the Pandas series we were talking about multi-indexed objects. But sometimes multi-indexed objects are created behind the scenes. The two common scenarios involve passing the following to the constructor of a Series or DataFrame object:

multiple index arrays
dictionaries where the keys are tuples containing the indices

Let’s have a look at them now.

MULTIPLE INDEX ARRAYS

Passing multiple (often two, but there may be more) index array is a quick way of creating a multi-indexed object. Hare’s an example:

In [2]:

import numpy as np
import pandas as pd

# Let's create a DataFrame with random numeric data.
# The arguments passed to the np.random.randint functions are: 
# 1000 - from value, inclusive
# 2000 - to value, exclusive
# (4, 4) - shape
a = pd.DataFrame(np.random.randint(1000, 2000, (4, 4)),
                 index=[['shirts', 'shirts', 'pants', 'pants'], ['blue', 'red', 'blue', 'red']],
                 columns=['spring', 'summer', 'fall', 'winter'])

# And here's our multi-indexed DataFrame object.
a

Out[2]:

		spring	summer	fall	winter
shirts	blue	1180	1438	1838	1507
shirts	red	1335	1604	1998	1845
pants	blue	1628	1329	1951	1174
pants	red	1345	1560	1587	1535

DICTIONARIES

The other method of implicitly creating multi-indexed objects is by means of dictionaries. We just have to pass the indices in tuples as keys. Here’s an example:

In [3]:

# This time we'll create a Series object.
# Let's define the dictionary with our data.
data = {('product 1', 'regular price'): 125,
        ('product 1', 'reduced price'): 84,
        ('product 2', 'regular price'): 265,
        ('product 2', 'reduced price'): 182,
        ('product 3', 'regular price'): 350,
        ('product 3', 'reduced price'): 280}

# Let's use the dictionary in the constructor of a Series object.
b = pd.Series(data)

# Here's our multi-indexed object.
b

Out[3]:

product 1  regular price    125
           reduced price     84
product 2  regular price    265
           reduced price    182
product 3  regular price    350
           reduced price    280
dtype: int64

In the next article we’ll see how to explicitly create multi-indexed objects.