Some people find the course interesting and would like to find out more about how python can be used for data analysis and how to continue learning about Python for (Geographic) Data Science. This section collects a few references to point you in that direction.
Serge Rey, Dani Arribas-Bel and Levi Wolf are currently working on an open book for Geographic Data Science in Python:
https://geographicdata.science/book/). Note that the book is still currently being developed so, although there is a portion of if that is fairly advanced, there are some chapters that are still not available. More information about the project is available in the project’s website https://geographicdata.science).Two excellent (free) books that will introduce you to Python and statistics are:
http://greenteapress.com/wp/think-python)
by Allen B. Downey,
which introduces the reader to the “Python way of thinking” and has a
fantastic pedagogy behind it.http://greenteapress.com/wp/think-stats-2e/),
also by Allen B. Downey.
This book is more geared towards the use for Python in a statistical
context. Mind, this is a book about statistics, not about Python, but all
the examples and computational illustrations, which are core to the author’s
approach to teaching the subject, are in Python.There are two popular books that are good introductions to using Python for Data Science:
https://www.oreilly.com/library/view/python-for-data/9781491957653/),
by Wes McKinney, the creator of the pandas library. This an excellent
comprehensive introduction to using Python to access, manipulate, transform
and begin to explore your data. There is very limited treatment of any
statistical methodology but, remember, 80% of the time spent analyzing data
is on this type of operations rather than advanced modelling!http://shop.oreilly.com/product/0636920034919.do),
by Jake VanderPlas. This is a cookbook so, rather than being structured
around a coherent narrative, it is made up of “recipes” for specific tasks
that are common in Data Science. There is a bit of overlap with “Python for
Data Analysis” but this one includes more content on machine learning
techniques.Note these are introductions to Data Science, and do not necessarily have a geographic component.
Several of the core packages for data science written in Python have prepared impressive resources as part of their documentation. In some cases, these references are a fantastic summary not only of the functionality of the library, but also of the methods behind them. Here is a short list with some of the most relevant ones:
pandas
(http://pandas.pydata.org/pandas-docs/stable/):
the pandas documentation contains very throrough tutorials on every
detail about how the library is built and how it can be used in a large
variety of contexts. Maybe not the best to follow from start to finish, but
the ideal place to understand how common operations such as table joins or
group-by operations work.seaborn
(http://seaborn.pydata.org/tutorial.html: its tutorial
page contains a detailed account of all the functionality provided to
satisfy your plotting needs without the difficulty of having to go always to
matplotlib. In addition, it includes fantastic tutorials for concepts
relevant to visualisation tasks, from how to choose an appropriate color
palette to how to visualise categorical data.scikit-learn
(http://scikit-learn.org/stable/) is the
main entry gate to the world-class machine learning library. In addition to
a detailed description of functionality, there are several tutorials that
are fantastic introductions to the methods behind this tool. scikit-learn
incorporates state-of-the-art machine learning and is constantly updated,
along with its website.statsmodels
(http://www.statsmodels.org/stable/index.html)
is the main library for statistical modelling in Python. It is not up to par
with what one could find in R, but it incorporates a large amount of
foundational techniques, from generalised linear models to hierarchical
models or time series analysis. The library website contains reference to
functionality and several useful tutorials.geopandas
(http://geopandas.org/) contains the best
one-stop-shop for learning how to read, manipulate and transform spatial
data in Python. The library itself is an abstraction of several other more
low-level libraries that acts as the glue to make working with all of the as
seamless as possible. The website includes tutorials on file formats,
projections, and spatial operations, among other topics.pysal (http://pysal.org/) is
the best resource to explore functionality in Python to perform spatial
analysis. The library is currently going under an overhaul that will mean
more related packages will be available through the PySAL “brand” so, its
website represents the best one-stop shop to see what is possible in terms
of spatial analysis in Python.http://www.scipy-lectures.org/) is fairly
low-level, so maybe not for absolute beginners but, if you need to use
Python to write your own application, this might be a great starting place.There are several online resources to learn Python, to a point that it becomes too hard to know where to start. One of the most useful ones is the recordings of the SciPy conference for scientific computing in Python, which takes place every July in Austin (TX). The conference includes both short presentations of new packages and projects and 4h workshops that delve into the details of the main functionality in the scientific Python stack. The playlist of the 2018 edition is available at:
https://www.youtube.com/playlist?list=PLYx7XA2nY5Gd-tNhm79CNMe_qvi35PgUR
And, in the context of Geographic Data Science, the following two workshops are of particular interest:
In addition to disconnected resources, there are starting to appear full bodied courses on the use of Python for (Geographic) Data Science. Here are a couple of interest:
http://darribas.org/gds19/) is a course
taught entirely in Python at the University of Liverpool by Dani Arribas-Bel.https://www.datacamp.com/courses/working-with-geospatial-data-in-python), by Joris Van den Bossche and Dani Arribas-Bel is a short online course on the DataCamp platform (paywalled).