Data science emerged as a multi-disciplinary field aimed at creating value from data. This course starts with an overview of data science and related fields and then defines data types and sources. Special focus is put on the assessment of data quality and electronic data processing.

Use of data-driven methods has become vital for businesses, and this course outlines how data-driven approaches can be integrated within a business context and how operational decisions can be made using data-driven methods.

Finally, this course highlights the importance of statistics and machine learning in the field of data science and gives an overview of relevant methods and approaches.

1. Introduction to Data Science
   1. Definition of the term „data science“
   2. Data science and related fields
   3. Data science activities
2. Data
   1. Data types and data sources
   2. The 5Vs of data
   3. Data curation and data quality
   4. Data engineering
3. Data Science in Business
   1. Identification of use cases
   2. Performance evaluation
   3. Data-driven operational decisions
   4. Cognitive biases
4. Statistics
   1. Importance of statistics for data science
   2. Important statistical concepts
5. Machine Learning
   1. Role of machine learning in data science
   2. Overview of machine learning approaches

The goal of data science can be summarized as the extraction of insights (hence, value) from data. It is self-evident that this objective cannot be successfully achieved based on unreliable and untrustworthy data. This course aims at establishing the notion of data quality and the pertinent methods for data quality management. Furthermore, techniques for acquiring data as well as formatting and tidying data in order to make it suitable for subsequent analytical treatment are covered.

1. Data Quality
   1. Introduction to data quality
   2. Data quality dimensions and issue types
2. Data Quality Management
   1. Data governance and stewardship
   2. Activities and processes
3. Data Acquisition
   1. Web scraping
   2. Data APIs
4. Working with Common Data Formats
   1. Text-based formats (CSV, XML, JSON)
   2. Binary formats (HDF 5, Parquet, Arrow)
5. Tidy Data
   1. Structuring
   2. Cleansing
   3. Enrichment

A core part of data science is creating value from data. This means not only the creation of sophisticated predictive models but also the development of these models according to modern software development principles.

This course gives a detailed overview of the relevant methods and paradigms which data scientists need to know in order to develop enterprise-grade models.

This course discusses traditional and agile project management techniques, highlighting both the Kanban and Scrum approaches. It explores relevant software development paradigms such as test-driven development, pair programming, mob programming, and extreme programming.

Special focus is given to the topic of testing and the consideration of how to bring a model into a production environment.

1. Traditional Project Management
   1. Requirements engineering
   2. Waterfall model
   3. Rational unified process
2. Agile Project Management
   1. Criticism of the waterfall model
   2. Introduction to SCRUM
   3. Introduction to Kanban
3. Testing
   1. Why testing?
   2. Unit trests
   3. Integration tests
   4. Performance monitoring
4. Software Development Paradigms
   1. Test-driven development (TDD)
   2. Pair programming
   3. Mob programming
   4. Extreme programming
5. From Model to Production
   1. Continuous delivery
   2. Continuous integration
   3. Building a scalable environment

This project course will give students hands-on experience in the challenging task of bringing a predictive model into a production environment. Students will need to consider practical aspects such as data storage and processing, as well as constraints such as service availability and the maximum amount of time a model is allowed to run due to external project requirements.

Through this course, students will obtain holistic overview of the integration of predictive models into enterprise-grade applications or services.

• This project course focuses on practical aspects of ensuring that a predictive model can run in a production environment. The students start with a chosen use case and model and then evaluate the requirements which need to be fulfilled so that the model can be used as part of an enterprise application or app. Students need to evaluate requirements in terms of data storage, processing and throughput, and availability of the service, as well as the persistency, serving, and versioning of the model itself. Monitoring the execution of model predictions and raising alerts in cases of operational issues is a core part of building a reliable model pipeline. All relevant artifacts and considerations are documented by the students in a project report.

Data are often considered the “new oil”, the raw material from which value is created. To harness the power of data, the data need to be stored and processed on a technical level. This course introduces the four “Vs” of data, as well as typical data sources and types.

The course discusses the most common data storage formats encountered in modern systems, focusing both on text-based as well as binary data formats.

Handling large amounts of data poses significant challenges for the underlying infrastructure. The course discusses the most important distributed and streaming data handling frameworks which are used in leading edge applications.

1. Data Types and Data Sources
   1. The 4Vs of data: volume, velocity, variety, veracity
   2. Data sources
   3. Data types
2. Text-Based and Binary Data Formats
   1. Simple formats: CSV, YAML
   2. XML
   3. JSON
   4. Hierarchical data format 5 (HDF 5)
   5. Apache Parquet
   6. Apache Arrow
3. NoSQL data stores
   1. Introduction and motivation
   2. Approaches and technical concepts
4. Distributed Systems
   1. Hadoop & MapReduce
   2. Hadoop file system (HDFS)
   3. Spark
   4. DASK
5. Streaming Frameworks
   1. Spark streaming
   2. Kafka

Many of the recent advances in data science, particularly machine learning and artificial intelligence, rely on comprehensive data storage and computing power. Cloud computing is one way of providing that power in a scalable way, without considerable upfront investment in hardware and software resources. This course introduces the area of cloud computing together with its enabling technologies. Moreover, the most cutting-edge advances like serverless computing and storage are illustrated. Finally, a thorough overview on popular cloud offerings, especially in regard to analytics capabilities, is given.

1. Introduction to Cloud Computing
   1. Fundamentals of Cloud computing
   2. Cloud Service Models
   3. Benefits and Risks
2. Enabling Technology
   1. Virtualization and Containerization
   2. Storage Technology
   3. Networks and RESTful Services
3. Serverless Computing
   1. Introduction to Serverless Computing
   2. Benefits
   3. Limitations
4. Established Cloud Platforms
   1. General Overview
   2. Google Cloud Platform
   3. Amazon Web Services
   4. Microsoft Azure
   5. Platform Comparison
5. Data Science in the Cloud
   1. Provider-independent services and tools
   2. Google Data Science and Machine Learning Services
   3. Amazon Web Services Data Science and Machine Learning Services
   4. Microsoft Azure Data Science and Machine Learning Services

This course explores concepts of data engineering. Data engineering is concerned with the infrastructure aspects of data science such as data storage and provision, as well as the provisioning of suitable operational environments.

After laying out foundational notions and concepts of the discipline, this course addresses important developments in storage technology; aspects of systems architecture for processing data at scale; containerization as a modern take on virtualization; and the logic of data pipelines and associated operational aspects. Important issues pertaining to data security and protection are also given appropriate attention.

1. Foundations of Data Engineering
   1. Reliability
   2. Scalability
   3. Maintainability
2. NoSQL In Depth
   1. Fundamentals of NoSQL
   2. Established NoSQL solutions
3. Architectures for Data Processing at Scale
   1. Batch processing architectures
   2. Architectures for stream and complex event processing
   3. Lambda architecture
4. Containerization In Depth
   1. Docker containers
   2. Container management
5. Governance & Security
   1. Data protection
   2. Data security
   3. Data governance
6. Operational Aspects
   1. Defining principles of DataOps
   2. Building and maintaining data pipelines
   3. Metrics and monitoring

The focus of this course is the implementation of a real-world data engineering use case in the form of a student portfolio.

To this end, students choose a project subject from the various sub-domains of data engineering. Examples include setting up a Docker container environment or dockerized service; implementing a data pipeline according to DataOps principles; and setting up an NoSQL data store.

The goal is for students to demonstrate they can transfer theoretical knowledge to an implementation scenario that closely mimics practical work in a professional data engineering setting.

• This course covers the practical implementation of approaches and techniques covered in the preceding methodological course in a project-oriented setting. Each participant must produce a portfolio detailing and documenting the work. Porfolio themes are chosen from a list, or suggested by the students in accord with the tutor.