[Pycon] [new paper] "Pietro Battiston" - You don't need n dimensions when you have pandas

[info a pycon.it]

Title: You don't need n dimensions when you have pandas
Duration: 240 (includes Q&A)
Q&A Session: 0
Language: en
Type: Talk

Abstract: This training, of introductory level, will be composed of two parts.

Parte I : introduction to pandas
------------------------------------

The first part will consist in a "classic" introduction to ``pandas`` (passing through a quick exposition of ``numpy``): how it works, what it provides, what is vectorization.

We will see what is an index and why it is important, we will look at some "standard" data types and see how the new ``ExtensionArray`` provide us with definitely non-standard data types.

Part II : a table, many dimensions
------------------------------------

In the second part, we will focus on particular types of indexes, ``MultiIndex``es, and on the many ``pandas`` features allowing us to use them to store and analyze multidimensional data. We will discover together _why_ ``pandas`` does not have (any more) data structures with more than 2 dimensions, and why we should not regret them. We will then look at further ways to "restructure" data, including ``groupby`` e ``window``.

Tags: [u'pydata', u'data-structures', u'data-analysis', u'pandas', u'numpy']