Professur Numerische Mathematik | Fakultät für Mathematik

Introduction to Data Science (4V, 2Ü) Prof. Ernst, WS 2018/19

Content

Tentative list of course topics:

Introduction: What is Data Science
Learning Theory
Regression
Neural Networks
Classification
Clustering and Tree-Based Methods
Support Vectors
Unsupervised Learning

Notices

Extra lab session	In place of the lecture on Thursday, December 20, there will be an extra lab session during the regular class time in the computer pool.
Class cancellations	There will be no class on Monday, December 17.
Rescheduled lecture	To make up for the lecture on December 10 lost to the railway strike, we will have a lecture in place of the lab session 13:35-15:05 on Tuesday, December 11 in computer pool.
Temporary lecture room change	On Thursday, November 1 and Monday, November 5, the lecture will take place in Room 2/N101.
New time for Lab Exercises	Beginning on Tuesday, October 23, 2018, we will start our labs 10 minutes earlier, i.e., 13:35.
Cancellation	There will be no lab exercise session on Tuesday, October 16.
Note	To participate in the lab exercises, all students should have an account with the MRZ (Mathematics Computing Center). Those who do not already have one, please apply for one by following this page and collect your login credentials with Ms. Margit Matt (Rh39, Room 704).
First Lab Exercises	Tuesday, October 9, 2018.
First Class	Monday, October 8, 2018.

Listing of this course in the electronic Vorlesungsverzeichnis (course directory):

Keine Lehrveranstaltung gefunden.

Lecture

Literature

James, Witten, Hastie & Tibshirani. An Introduction to Statistical Learning – with Applications in R. Springer 2013. Available online at this page.
Here's a continually updated annotated reading list for the course (16.01.2019).

Slides

What is Data Science? (05.02.2019)
Learning Theory (05.02.2019)
Linear Regression (05.02.2019)
Classification (05.02.2019)
Resampling Methods (05.02.2019)
Linear Model Selection and Regularization (05.02.2019)
Nonlinear Regression Models (05.02.2019)
Tree-Based Methods (05.02.2019)
Support Vector Machines (05.02.2019)
Unsupervised Learning (05.02.2019)
All slides (05.02.2019)

Exercises

Installation of Programming Environment under Linux (64 bit)

If you want to do the homework on your personal computers, you may clone the programming environment used in the labs. Get miniconda from this web page and follow the steps in the installation dialogue. Next, download the specification file spec-file.txt used in the labs and create a conda environment (under Linux):

conda create --name DS2018 --file spec-file.txt

Installation of Programming Environment under Windows and MacOS

Download miniconda for your distribution by following this link and follow the installation instructions. Next, download the yml-file containing the packages used in the labs and create a conda environment in a miniconda/Anaconda shell:

conda env create -f DS2018.yml

If your plots are not displayed in the browser, this might be due to a missing package. After sourcing of the correct environment, the following might help in some cases python -m ipykernel install --user Please refer to Conda (Installation under Windows, Linux and MacOS) and Conda (Managing environments) for further information.

Material

In order to start the jupyter notebooks you have to open a terminal and source our conda environment DS2018 via

source /LOCAL/Software/DataScience2018/setup_env

Next, change the directory to your exercise folder and download the jupyter notebook (right click and "Save link as") into this folder. Finally, start the notebook via the command (make sure you see the (DS2018) in front of your username):

jupyter notebook Problem_Sheet_XX.ipynb