Creating a DataFrame from objects in pandas

This introduction to pandas is derived from Data School's pandas Q&A with my own notes and code.

Creating a pandas DataFrame from another object¶

In [1]:

import pandas as pd

In [35]:

# .DataFrame is a constructor

# create a dictionary
ex_dic = {
    'id': [100, 101, 102],
    'color': ['red', 'blue', 'red']
}

# create a list of strings
columns = ['id', 'color']

index = ['a', 'b', 'c']

# Passing a dictionary
# key: column name
# value: series of values
df = pd.DataFrame(ex_dict, columns=columns, index=index)
df

Out[35]:

	id	color
a	100	red
b	101	blue
c	102	red

In [16]:

# list of lists
list_of_lists = [
    [100, 'red'],
    [101, 'blue'],
    [102, 'red']
]

# columns
columns = ['id', 'color']

pd.DataFrame(list_of_lists, columns=columns)

Out[16]:

	id	color
0	100	red
1	101	blue
2	102	red

In [18]:

import numpy as np

In [20]:

# create 4x2 random array 
# array is a list of lists
arr = np.random.rand(4, 2)
arr

Out[20]:

array([[ 0.66867334,  0.0496808 ],
       [ 0.24225703,  0.17014163],
       [ 0.37133698,  0.3160525 ],
       [ 0.76333377,  0.54704594]])

In [23]:

# columns
columns_new = ['one', 'two']

# pass in array and columns
pd.DataFrame(arr, columns=columns_new)

Out[23]:

	one	two
0	0.668673	0.049681
1	0.242257	0.170142
2	0.371337	0.316052
3	0.763334	0.547046

In [25]:

# np.arange
# inclusive of start
# exclusive of end
# step: 1 
# ascending order

# np.random.randint
# 60 to 101
# any random integer

# key: column name
# value: series of values
dict_new = {
    'student': np.arange(100, 110, 1),
    'test': np.random.randint(60, 101, 10)
}

pd.DataFrame(dict_new)

Out[25]:

	student	test
0	100	66
1	101	68
2	102	74
3	103	73
4	104	92
5	105	100
6	106	65
7	107	73
8	108	100
9	109	82

In [27]:

# we can set the index
pd.DataFrame(dict_new).set_index('student')

Out[27]:

	test
student
100	66
101	68
102	74
103	73
104	92
105	100
106	65
107	73
108	100
109	82

In [33]:

# creating a series

lst = ['round', 'square']
ind = ['c', 'b']
series_name = 'shape'

s = pd.Series(lst, index=ind, name=series_name)
s

Out[33]:

c     round
b    square
Name: shape, dtype: object

In [36]:

df

Out[36]:

	id	color
a	100	red
b	101	blue
c	102	red

In [38]:

# concatenate columns
# auto-alignment
df2 = pd.concat([df, s], axis=1)
df2

Out[38]:

	id	color	shape
a	100	red	NaN
b	101	blue	square
c	102	red	round

In [45]:

# say you want to add a value 'rectangle'
df2.loc['a', 'shape'] = 'rectangle'

In [44]:

df2

Out[44]:

	id	color	shape
a	100	red	rectangle
b	101	blue	square
c	102	red	round

Tags: pandas