Pandas

Pandas

Maik Röder
Python Barcelona Meetup
7. February 2013

Python Consultant
maikroeder@gmail.com

Friday, February 8, 13

Pandas
• Powerful and productive Python data
analysis and management library
• Panel Data System
• Open Sourced by AQR Capital
Management, LLC in late 2009
• 30.000 lines of tested Python/Cython code
• Used in production in many companies

Pandas
• Rich data structures and functions to make
working with structured data fast, easy, and
expressive
• Built on top of Numpy with its high
performance array-computing features
• ﬂexible data manipulation capabilities of
spreadsheets and relational databases
• Sophisticated indexing functionality
• slice, dice, perform aggregations, select
subsets of data

The ideal tool for data
scientists
• Munging data
• Cleaning data
• Analyzing data
• Modeling data
• Organizing the results of the analysis into a
form suitable for plotting or tabular display


Series
• one-dimensional array-like object
>>> s = Series((1,2,3,4,5))
• Contains an array of data (of any Numpy
data type)
>>> s.values
• Has an associated array of data labels, the
index (Default index from 0 to N - 1)
>>> s.index

Series data structure
>>> import numpy
>>> randn = numpy.random.randn
>>> from pandas import *
>>> s = Series(randn(3),('a','b','c'))
>>> s
a -0.889880
b 1.102135
c -2.187296
>>> s.mean()
-0.65834710697853194


Series to/from dict
• Series to Python dict - No more explicit order
>>> dict(s)
{'a': -0.88988001423312313,
'c': -2.1872960440695666,
'b': 1.1021347373670938}
• Back to a Series with a new Index from sorted
dictionary keys
>>> Series(dict(s))
a -0.889880
b 1.102135
c -2.187296

Reindexing labels
>>> s
a -0.496848
b 0.607173
c -1.570596
>>> s.index
Index([a, b, c], dtype=object)
>>> s.reindex(['c','b','a'])
c -1.570596
b 0.607173
a -0.496848

Vectorization
>>> s + s
a -1.779760
b 2.204269
c -4.374592
• Series work with Numpy
>>> numpy.exp(s)
a 0.410705
b 3.010586
c 0.112220

DataFrame
• Like data.frame in the statistical
language/package R
• 2-dimensional tabular data structure
• Data manipulation with integrated
indexing
• Support heterogeneous columns
• Homogeneous columns

DataFrame
>>> d = {'one': s*s, 'two': s+s}
>>> DataFrame(d)
one two
a 0.791886 -1.779760
b 1.214701 2.204269
c 4.784264 -4.374592
>>> df.index
Index([a, b, c], dtype=object)
>>> df.columns
Index([one, two], dtype=objec)


Dataframe add column
• Add a third column
>>> df['three'] = s * 3
• It will share the existing index
>>> df
one two three
a 0.791886 -1.779760 -2.669640
b 1.214701 2.204269 3.306404
c 4.784264 -4.374592 -6.561888


Access to columns

• Access by attribute • Access by dict like
notation
>>> df.one >>> df['one']
one one
a 0.791886 a 0.791886
b 1.214701 b 1.214701
c 4.784264 c 4.784264


Reindexing

>>> df.reindex(['c','b','a'])
>>> df
one two three
c 4.784264 -4.374592 -6.561888
b 1.214701 2.204269 3.306404
a 0.791886 -1.779760 -2.669640


Drop entries from an axis

>>> df.drop('c')
b 1.214701 2.204269 3.306404
a 0.791886 -1.779760 -2.669640
>>> df.drop(['b,'a'])
one two three
c 4.784264 -4.374592 -6.561888


Descriptive statistics
>>> df.mean()
one 2.263617
two -1.316694
three -1.975041
• Also: count, sum, median, min,
max, abs, prod, std, var,
skew, kurt, quantile, cumsum,
cumprod, cummax, cummin


Computational Tools
• Covariance
>>> s1 = Series(randn(1000))
>>> s2 = Series(randn(1000))
>>> s1.cov(s2)
0.013973709323221539
• Also: pearson, kendall, spearman


This and much more...
• Group by: split-apply-combine
• Merge, join and aggregate
• Reshaping and Pivot Tables
• Time Series / Date functionality
• Plotting with matplotlib
• IO Tools (Text, CSV, HDF5, ...)
• Sparse data structures

Resources

• http://pypi.python.org/pypi/pandas
• http://code.google.com/p/pandas


Out now...


Pandas

More Related Content

What's hot

Viewers also liked

Similar to Pandas

More from maikroeder

Recently uploaded

Pandas