Blog 5 Introduction to Pandas DataFrame

In [176]:

#Adjusting the cell width and margin

In [184]:

from IPython.display import display, HTML

display(HTML(data="""
<style>
    div#notebook-container    { width: 100%; }
    div#notebook-container    { margin-left: -2.8%; }
    div#menubar-container     { width: 65%; }
    div#maintoolbar-container { width: 99%; }
</style>
"""))

In [148]:

# Importing the libraries

In [1]:

import pandas as pd
import numpy as np

In [ ]:

# Lets create a dummy data frame and
# review the functions
# associated with a data frame 

In [76]:

df_m = pd.DataFrame(np.array([[1,2,3],[10,11,12]]), columns=['col1','col2','col3'])
df_m

Out[76]:

	col1	col2	col3
0	1	2	3
1	10	11	12

In [120]:

# Number of rows and columns.The output is in the form
# of a tuple
# Two rows and 4 columns
df_m.shape

Out[120]:

(2, 4)

In [121]:

# Get number of rows
df_m.shape[0]

Out[121]:

2

In [122]:

# Get number of columns
df_m.shape[1]

Out[122]:

4

In [77]:

# Lets look at the top record
df_m.head(1)

Out[77]:

	col1	col2	col3
0	1	2	3

In [78]:

# Lets look at the bottom record
df_m.tail(1)

Out[78]:

	col1	col2	col3
1	10	11	12

In [81]:

# Change column names
df_m.columns=['Col3','Col4','Col5']
df_m

Out[81]:

	Col3	Col4	Col5
0	1	2	3
1	10	11	12

In [82]:

# Data Types of data frame
df_m.dtypes

Out[82]:

Col3    int32
Col4    int32
Col5    int32
dtype: object

In [ ]:

# Summarising the data using describe function

In [83]:

df_m.describe()

Out[83]:

	Col3	Col4	Col5
count	2.000000	2.000000	2.000000
mean	5.500000	6.500000	7.500000
std	6.363961	6.363961	6.363961
min	1.000000	2.000000	3.000000
25%	3.250000	4.250000	5.250000
50%	5.500000	6.500000	7.500000
75%	7.750000	8.750000	9.750000
max	10.000000	11.000000	12.000000

In [86]:

# Accessing the index of the data frame
list(df_m.index)

Out[86]:

[0, 1]

In [87]:

# Changing the index of the data frame
df_m.index = ['Ind1','Ind2']

In [18]:

df_m

Out[18]:

	Col3	Col4
Ind1	1	2
Ind2	3	4

In [88]:

# Accessing single column(results in a series)
# Series is the building block of dataframe
s1=df_m['Col3']
type(s1)

Out[88]:

pandas.core.series.Series

In [89]:

# Adding a data frame
df_m['Name']=['Roger','Nadal']
df_m

Out[89]:

	Col3	Col4	Col5	Name
Ind1	1	2	3	Roger
Ind2	10	11	12	Nadal

In [90]:

df_m.describe()
# Describe works only for Numeric columns

Out[90]:

	Col3	Col4	Col5
count	2.000000	2.000000	2.000000
mean	5.500000	6.500000	7.500000
std	6.363961	6.363961	6.363961
min	1.000000	2.000000	3.000000
25%	3.250000	4.250000	5.250000
50%	5.500000	6.500000	7.500000
75%	7.750000	8.750000	9.750000
max	10.000000	11.000000	12.000000

In [91]:

# Accessing Multiple column
df_m2=df_m[['Col3','Name']]
df_m2

Out[91]:

	Col3	Name
Ind1	1	Roger
Ind2	10	Nadal

In [92]:

# Returns a series
type(df_m['Name'])

Out[92]:

pandas.core.series.Series

In [93]:

# Returns a DataFrame
type(df_m[['Name']])

Out[93]:

pandas.core.frame.DataFrame

In [94]:

# Selecting rows using numerical values/slices
# The below selects data frame records 
# from 0 to 1 rows
df_m[:2]

Out[94]:

	Col3	Col4	Col5	Name
Ind1	1	2	3	Roger
Ind2	10	11	12	Nadal

In [95]:

# The below selects data frame records 
# from 0th row
df_m[:1]

Out[95]:

	Col3	Col4	Col5	Name
Ind1	1	2	3	Roger

In [97]:

# Filtering a dataset
pos1=list(df_m['Name']=="Roger")
pos1

Out[97]:

[True, False]

In [98]:

df_m[pos1]

Out[98]:

	Col3	Col4	Col5	Name
Ind1	1	2	3	Roger

In [99]:

# Filtering a dataset based on Multiple
# conditions
pos1=list(df_m['Name']=="Roger")
pos1

Out[99]:

[True, False]

In [100]:

pos2=list(df_m['Col3']==1)
pos2

Out[100]:

[True, False]

In [105]:

pos3=pos1 and pos2
pos3

Out[105]:

[True, False]

In [106]:

df_m[pos3]

Out[106]:

	Col3	Col4	Col5	Name
Ind1	1	2	3	Roger

In [ ]:

# Using Iloc for indexing

In [107]:

df_m.iloc[0:1]

Out[107]:

	Col3	Col4	Col5	Name
Ind1	1	2	3	Roger

In [108]:

df_m.iloc[0:2]

Out[108]:

	Col3	Col4	Col5	Name
Ind1	1	2	3	Roger
Ind2	10	11	12	Nadal

In [114]:

# Fist two rows and First Column
df_m.iloc[0:2,0:1]

Out[114]:

	Col3
Ind1	1
Ind2	10

In [115]:

# Fist two rows and First two Column
df_m.iloc[0:2,0:2]

Out[115]:

	Col3	Col4
Ind1	1	2
Ind2	10	11

In [116]:

# Fist two rows and Column at first
# index position
df_m.iloc[0:2,1:2]

Out[116]:

	Col4
Ind1	2
Ind2	11

In [118]:

# Fist two rows and all columns
df_m.iloc[0:2,0:len(df_m.columns)]

Out[118]:

	Col3	Col4	Col5	Name
Ind1	1	2	3	Roger
Ind2	10	11	12	Nadal

In [123]:

# All rows and all columns
df_m.iloc[0:df_m.shape[0],0:len(df_m.columns)]

Out[123]:

	Col3	Col4	Col5	Name
Ind1	1	2	3	Roger
Ind2	10	11	12	Nadal