ONLINE COURSE – Python for data science, machine learning, and scientific computing (PDMS02) This course will be delivered live https://www.psstatistics.com/course/python-for-data-science-machine-learning... Dates 4 - 8 May 2020 Time zone - UK (GMT) Please email oliverhooker@psstatistics.com with any questions Course Overview: Python is one of the most widely used and highly valued programming languages in the world, and is especially widely used in data science, machine learning, and in other scientific computing applications. This course provides both a general introduction to programming with Python and a comprehensive introduction to using Python for data science, machine learning, and scientific computing. The skills learnt will be of value to any marine a mammal researcher handling large data sets. The major topics that we will cover include the following: the fundamentals of general purpose programming in Python; using Jupyter notebooks as a reproducible interactive Python programming environment; numerical computing using numpy; data processing and manipulations using pandas; data visualization using matplotlib, seaborn, ggplot, bokeh, altair, etc; symbolic mathematics using sympy; data science and machine learning using scikit-learn, keras, and tensorflow; Bayesian modelling using PyMC3 and PyStan; high performance computing with Cython, Numba, IPyParallel, Dask. Overall, this course aims to provide a solid introduction to Python generally as a programming language, and to its principal tools for doing data science, machine learning, and scientific computing. (Note that this course will focus on Python 3 exclusively given that Python 2 has now reached it end of life). Monday 4th – Classes from 09:30 to 17:30 • Topic 1: The What and Why of Python. In order to provide some general background and context, we will describe Python where came from, what its major design principles and intended use was originally, and where and how it is now currently used. We will see that Python is now extremely widely used, especially in powering the web, in data science and machine learning, and system level programming. Here, we also compare and contrast Python and R, given that both are extremely widely used in data science. • Topic 2: Installing and setting up Python. There are many ways to write and execute code in Python. Which to use depends on personal preference and the type of programming that is being done. Here, we will explore some of the commonly used Integrated Development Environments (IDE) for Python, which include Spyder and PyCharm. Here, we will also mention and briefly describe Jupyter notebooks, which are widely used for scientific applications of Python, and are an excellent tool for doing reproducible interactive work. We will cover Jupyter more extensively starting on Day 3. Also as part of this topic, we will describe how to use virtual environments and package installers such as pip and conda. • Topic 3: Introduction to Python: Data Structures. We will begin our coverage of programming with Python by introducing its different data structures.and operations on data structures This will begin with the elementary data types such as integers, floats, Booleans, and strings, and the common operations that can be applied to these data types. We will then proceed to the so-called collection data structures, which primarily include lists, dictionaries, tuples, and sets. • Topic 4: Introduction to Python: Programming. Having introduced Python’s data types, we will now turn to how to program in Python. We will begin with iteration, such as the for and while loops. We will then cover conditionals and functions. Tuesday 5th – Classes from 09:30 to 17:30 • Topic 5: Modules, packages, and imports. Python is extended by hundreds of thousands of additional packages. Here, we will cover how to install and import these packages, and also how to write our own modules and packages. • Topic 6: Numerical programming with numpy. Although not part of Python’s official standard library, the numpy package is the part of the de facto standard library for any scientific and numerical programming. Here we will introduce numpy, especially numpy arrays and their built in functions (i.e. “methods”). • Topic 7: Data processing with pandas. The pandas library provides means to represent and manipulate data frames. Like numpy, pandas can be see as part of the de facto standard library for data oriented uses of Python. • Topic 8: Object Oriented Programming. Python is an object oriented language and object oriented programming in Python is extensively used in anything beyond the very simplest types of programs. Moreover, compared to other languages, object oriented programming in Python is relatively easy to learn. Here, we provide a comprehensive introduction to object oriented programming in Python. • Topic 9: Other Python programming features. In this section, we will cover some important features of Python not yet covered. These include exception handling, list and dictionary comprehensions, itertools, advanced collection types including defaultdict, anonymous functions, decorators, etc. Wednesday 6th – Classes from 09:30 to 17:30 • Topic 10: Jupyter notebooks and Jupyterlab. Although we have already introduced Jupyter notebooks, here we will explore them properly. Jupyter notebooks are reproducible and interactive computing environment that support numerous programming languages, although Python remains the principal language used in Jupyter notebooks. Here, we’ll explore their major features and how they can be shared easily using GitHub and Binder. • Topic 11: Data Visualization. Python provides many options for data visualization. The matplotlib library is a low level plotting library that allows for considerable control of the plot, albeit at the price of a considerable amount of low level code. Based on matplotlib, and providing a much higher level interface to the plot, is the seaborn library. This allows us to produce complex data visualizations with a minimal amount of code. Similar to seaborn is ggplot, which is a direct port of the widely used R based visualization library. In this section, we will also consider a set of other visualization libraries for Python. These include plotly, bokeh, and altair. • Topic 12: Symbolic mathematics. Symbolic mathematics systems, also known as computer algebra systems, allow us to algebraically manipulate and solve symbolic mathematical expression. In Python, the principal symbolic mathematics library is sympy. This allows us simplify mathematical expressions, compute derivatives, integrals, and limits, solve equations, algebraically manipulate matrices, and more. • Topic 13: Statistical data analysis. In this section, we will describe how to perform widely used statistical analysis in Python. Here we will start with the statsmodels package, which provides linear and generalized linear models as well as many other widely used statistical models. We will also introduce the scikit-learn package, which we will more widely use on Day 4, and use it for regression and classification analysis. Thursday 7th – Classes from 09:30 to 17:30 • Topic 14: Machine learning. Python is arguably the most widely used language for machine learning. In this section, we will explore some of the major Python machine learning tools that are part of the scikit-learn package. This section continues our coverage of this package that began in Topic 12 on Day 3. Here, we will cover machine learning tools such as support vector machines, decision trees, random forests, k-means clustering, dimensionality reduction, model evaluation, and cross-validation. • Topic 15: Neural networks and deep learning. A popular subfield of machine learning involves the use of artificial neural networks and deep learning methods. In this section, we will explore neural networks and deep learning using the keras library, which is a high level interface to neural network and deep learning libraries such as Tensorflow, Theano, or the Microsoft Cognitive Toolkit (CNTK). Examples that we will consider here include image classification and other classification problems taken from, for example, the UCI Machine Learning Repository. Friday 8th – Classes from 09:30 to 16:00 • Topic 16: Bayesian models. Two probabilistic programming languages for Bayesian modelling in Python are PyMC3 and PyStan. PyMC3 is a Python native probabilistic programming language, while PyStan is the Python interface to the Stan programming language, which is also very widely used in R. Both PyMC3 and PyStan are extremely powerful tools and can implement arbitrary probabilistic models. Here, we will not have time to explore either in depth, but will be able to work through a number of nontrivial examples, which will illustrate the general feature and usage of both languages. • Topic 17: High performance Python. The final topic that we will consider in this course is high performance computing with Python. While many of the tools that we considered above extremely quickly because they interface with compiled code written in C/C++ or Fortran, Python itself is a high level dynamically typed and interpreted programming language. As such, native Python code does not execute as fast as compiled languages such as C/C++ or Fortran. However, it is possible to achieve compiled language speeds in Python by compiling Python code. Here, we will consider Cython and Numba, both of which allow us achieve C/C++ speeds in Python with minimal extensions to our code. Also, in this section, we will consider parallelization in Python, in particular using IPyParallel and Dask, both of which allow easy parallel and distributed processing using Python. -- Oliver Hooker PhD. PS statistics 2020 publications; Parallelism in eco-morphology and gene expression despite variable evolutionary and genomic backgrounds in a Holarctic fish. PLOS GENETICS (2020). IN PRESS www.PSstatistics.com facebook.com/PSstatistics/ twitter.com/PSstatistics 53 Morrison Street Glasgow G5 8LB +44 (0) 7966500340