This is the handout and homework repository for the course "ME 597 Data Analytics for Scientists and Engineers" which is currently taught (Fall 2020) by Prof. Ilias Bilionis at Purdue University. The course is fully online with the videos being accessible through EdX. Currently only registered Purdue students have access to the class. Hopefully, we will soon be able make the content available to everyone.

This course evolved from the ME 597/MA 598 "Introduction to Uncertainty Quantification" taught three times by Prof. Bilionis (the first time, Spring 2016 it was co-taught with Prof. Guang Lin). If you are interested in accessing the old versions of the course, they can be found here. Note, that there is also a, 1-credit, undergraduate version of the course under the name ME 297 "Introduction to Data Science for Mechanical Engineers." This version can be found here.

The material is publised under the GNU General Public License. You can reuse it in your own courses as soon as you also include the same lience and cite this repository. Please send me an email if you do as I would love to know!

• Python Basics Activity 1 (Python as a calculator)
• Python Basics Activity 2 (Python variables and types)
• Python Basics Activity 3 (Basic tuples and lists)
• Python Basics Activity 4 (Basics of strings and printing)
• Python Basics Activity 5 (Conditionals)
• Python Basics Activity 6 (Loops)
• Python Basics Activity 7 (Functions)
• Python Basics Activity 8 (Classes)
• Python Basics Activity 9 (Numpy arrays and linear algebra)
• Python Basics Activity 10 (The Python data analysis library)
• Python Basics Activity 11 (Basic plotting)

Below, I provide links that open up directly on Google Colab. If you want to view the Jupyter notebooks locally, please see the section named Running the notebooks on your personal computer.

• Lecture 1 - Introduction to Predictive Modeling

  • Reading Activity 1
  • Hands-on Activity 1.1 (The uncertainty propagation probability problem)
  • Hands-on Activity 1.2 (The model calibration problem)

• Lecture 2 - Basics of Probability Theory

  • Reading Activity 2
  • Hands-on Activity 2.1 (Experiment with the "randomness" of a coin toss)

• Lecture 3 - Discrete Random Variables

  • Reading Activity 3
  • Hands-on Activity 3.1 (Discrete random variables)

• Lecture 4 - Continuous Random variables

  • Reading Activity 4
  • Hands-on Activity 4.1 (The uniform distribution)
  • Hands-on Activity 4.2 (The Gaussian distribution)

• Lecture 5 - Collections of Random variables

  • Reading Activity 5
  • Hands-on Activity 5.1 (Collections of random variables)

• Lecture 6 - Random Vectors

  • Reading Activity 6
  • Hands-on Activity 6.1 (The multivariate normal - diagonal covariance case)
  • Hands-on Activity 6.2 (The multivariate normal - full covariance case)
  • Hands-on Activity 6.3 (The multivariate normal - marginalization)
  • Hands-on Activity 6.4 (The multivariate normal - conditioning)

• Lecture 7 - Basic sampling

  • Hands-on Activity 7.1 (Pseudo-random number generators)
  • Hands-on Activity 7.2 (Sampling the uniform)
  • Hands-on Activity 7.3 (Sampling the categorical)
  • Hands-on Activity 7.4 (Inverse sampling)

• Lecture 8 - The Monte Carlo Method for Estimating Expectations

  • Hands-on Activity 8.1 (The law of large numbers)
  • Hands-on Activity 8.2 (Estimating the variance)

• Lecture 9 - Monte Carlo Estimates of Various Statistics

  • Hands-on Activity 9.1 (Estimating the cumulative distribution function)
  • Hands-on Activity 9.2 (Estimating the probability density function via histograms)
  • Hands-on Activity 9.3 (Estimating predictive quantiles)
  • Hands-on Activity 9.4 (Application – Uncertainty propagation through an initial value problem)

• Lecture 10 - Quantify Uncertainty in Monte Carlo Estimates

  • Hands-on Activity 10.1 (Visualizing Monte Carlo uncertainty)
  • Hands-on Activity 10.2 (The central limit theorem)
  • Hands-on Activity 10.3 (Epistemic uncertainty of Monte Carlo estimates)
  • Hands-on Activity 10.4 (Application – Uncertainty propagation through a boundary value problem)

• Lecture 11 - Selecting Prior Information

  • Reading Activity 11
  • Hands-on Activity 11.1 (Information entropy)
  • Hands-on Activity 11.2 (The principle of maximum entropy for discrete random variables)
  • Hands-on Activity 11.3 (The principle of maximum entropy for continuous random variables)

• Lecture 12 - Analytical Examples of Bayesian Inference

  • Reading Activity 12
  • Hands-on Activity 12.1 (Bayesian parameter estimation)
  • Hands-on Activity 12.2 (Credible intervals)
  • Hands-on Activity 12.3 (Decision-making)
  • Hands-on Activity 12.4 (Predictive checking)

• Lecture 13 - Linear Regression via Least Squares

  • Reading Activity 13
  • Hands-on Activity 13.1 (Linear regression with a single variable)
  • Hands-on Activity 13.2 (Polynomial regression)
  • Hands-on Activity 13.3 (The generalized linear model)
  • Hands-on Activity 13.4 (Measures of predictive accuracy)

• Lecture 14 - Bayesian Linear regression

  • Reading Activity 14
  • Hands-on Activity 14.1 (Probabilistic interpretation of least squares – Estimating the measurement noise)
  • Hands-on Activity 14.2 (Maximum a posteriori estimate – Avoiding overfitting)
  • Hands-on Activity 14.3 (Bayesian linear regression)
  • Hands-on Activity 14.4 (The point-predictive distribution – Separating epistemic and aleatory uncertainty)

• Lecture 15 - Advanced Topics in Bayesian Linear regression

  • Reading Activity 15
  • Hands-on Activity 15.1 (The evidence approximation)
  • Hands-on Activity 15.2 (Automatic relevance determination)
  • Hands-on Activity 15.3 (Diagnostics for posterior predictive)

• Lecture 16 - Classification

  • Reading Activity 16
  • Hands-on Activity 16.1 (Logistic regression with a single variable)
  • Hands-on Activity 16.2 (Logistic regression with many features)
  • Hands-on Activity 16.3 (Making decisions)
  • Hands-on Activity 16.4 (Diagnostics for classification)
  • Hands-on Activity 16.5 (Multi-class logistic regression)

• Lecture 17 - Clustering and Density Estimation

  • Reading Activity 17
  • Hands-on Activity 17.1 (Clustering using k-means)
  • Hands-on Activity 17.2 (Density estimation using Gaussian mixture)

• Lecture 18 - Dimensionality Reduction

  • Reading Activity 18
  • Hands-on Activity 18.1 (Principal component analysis)
  • Hands-on Activity 18.2 (Clustering high-dimensional data)
  • Hands-on Activity 18.3 (Density estimation with high-dimensional data)

• Lecture 19 - State Space Models - Filtering Basics

  • Reading Activity 19
  • Hands-on Activity 19.1 (Object tracking example)

• Lecture 20 - State Space Models - Kalman Filters

  • Reading Activity 20
  • Hands-on Activity 20.1 (Kalman filters example)

• Lecture 21 - Gaussian Process Regression - Priors on Function Spaces

  • Reading Activities 21
  • Hands-on Activity 21.1 (Priors on function spaces)

• Lecture 22 - Gaussian Process Regression - Conditioning on Data

  • Reading Activity 22
  • Hands-on Activity 22.1 (Gaussian process regression without measurement noise)
  • Hands-on Activity 22.2 (Gaussian process regression with measurement noise)
  • • Hands-on Activity 22.3 (Tuning the hyperparameters)
  • • Hands-on Activity 22.4 (Multivariate regression and automatic relevance determination)

• Lecture 23 - Bayesian Global Optimization

  • Reading Activity 23
  • Hands-on Activity 23.1 (Maximum mean: A very bad information acquisition function)
  • Hands-on Activity 23.2 (Maximum upper interval)
  • Hands-on Activity 23.3 (Probability of improvement – No observation noise)
  • Hands-on Activity 23.4 (Expected improvement – No observation noise)
  • Hands-on Activity 23.5 (Expected improvement – With observation noise)
  • Hands-on Activity 23.6 (Quantifying epistemic uncertainty in the location and value of maximum)

• Lecture 24 - Deep Neural Networks

  • Reading Activity 24
  • Hands-on Activity 24.1 (The stochastic gradient descent algorithm and variants)

• Lecture 25 - Deep Neural Networks Continued

  • Reading Activity 25
  • Hands-on Activity 25.1 (Image classification)

• Lecture 26 - Physics-informed Deep Neural Networks

  • Reading Activity 26
  • Hands-on Activity 26.1 (Physics-informed regularization: Solving ODEs)
  • Hands-on Activity 26.2 (Physics-informed regularization: Solving PDEs)
  • Hands-on Activity 26.3 (Physics-informed regularization: Solving high-dimensional stochastic PDEs)

• Lecture 27 - Sampling Methods

  • Reading Activity 27
  • Hands-on Activity 27.1 (Introduction to probabilistic programming)
  • Hands-on Activity 27.2 (Example: Sampling from the Exponential with random walk Metropolis)
  • Hands-on Activity 27.3 (The Metropolis-Hastings algorithm)
  • Hands-on Activity 27.4 (Gibbs sampling)
  • Hands-on Activity 27.5 (Sequential Monte Carlo)

• Lecture 28 - Variational Inference

  • Reading Activity 28
  • Hands-on Activity 28.1 (Variational inference)

• Homework 1
• Homework 2
• Homework 3
• Homework 4
• Homework 5
• Homework 6
• Homework 7
• Homework 8
• Homework 9
• Homework 10

Make sure you have a Google account before you start. Then, you just click on the links above.

Find and download the right version of Anaconda for Python 3.7 from Continuum Analytics. This package contains most of the software we are going to need. Note: You do need Python 3 and note Python 2. The notebooks will not work with Python 2.

• We need C, C++, Fortran compilers, as well as the Python sources. Start the command line by opening "Anaconda Prompt" from the start menu. In the command line type:

conda config --append channels https://repo.continuum.io/pkgs/free
conda install mingw libpython

• Finally, you need git. As you install it, make sure to indicate that you want to use "Git from the command line and also from 3rd party software".

• Download and install the latest version of Xcode.

If you are using Linux, I am sure that you can figure it out on your own.

Independently of the operating system, use the command line to install the following Python packages:

• Seaborn, for beautiful graphics:

conda install seaborn

• PyMC3 for MCMC sampling:

conda install pymc3

• GPy for Gaussian process regression:

pip install GPy

• pydoe for generating experimental designs:

pip install pydoe

• fipy for solving partial differential equations using the finite volume method:

pip install fipy

*** Windows Users ***

You may receive the error

ModuleNotFoundError: No module named 'future'

If so, please install future and then install fipy:

pip install future

• scikit-learn for some standard machine learning algorithms implemented in Python:

conda install scikit-learn

• graphviz for visualizing probabilistic graphical models:

pip install graphviz

• Open the command line.
• cd to your favorite folder.
• Then, type:

git clone https://github.com/PredictiveScienceLab/data-analytics-se.git

• This will download the contents of this repository in a folder called data-analytics-se.
• Enter the uq-course folder:

cd data-analytics-se

• Start the jupyter notebook by typing the command:

jupyter notebook

• Use the browser to navigate the course, experiment with code etc.
• If the course content has been updated, type the following command (while being inside data-analytics-se) to get the latest version:

git pull origin master

Keep in mind, that if you have made local changes to the repository, you may have to commit them before moving on.