I part-time studied Data Science (M.Sc.) in 2021 — here is what it’s like
I started a Data Science Master back in 2018 at the time the Data Scientist Hype almost reached its height.
If you go back years ago, the shiny titles in the early 2010s were programmers and web designers. The salaries for the two were great back then, but have plateaued since as supply caught up with demand.
There is a reason Data Scientist is in the top 3 for job rankings, and it’s because their demand is absolutely ridiculous and in no sight of slowing down.
Where Does This Demand Come From?
Data-driven decision making. That is the simple answer to this question. To be a successful company in the 21st century you have to use data to your advantage.
Before many were doing this by using excel to analyze data, but now anyone can have access to and use data-crunching tools like:
- Google Analytics — Digital marketing cloud-based service
- Tableau, Power BI — Data visualization tools for business intelligence
- Python, R — Programming languages used to perform complicated analysis with a few lines of code
I studied Data Science in the Master’s program and would like to reveal the most important findings in this article. First of all, I will talk about my decision — why I chose to study Data Science. Afterwards, I will present the contents before outlining the advantages and disadvantages of studying Data Science.
With the exponential growth of Big Data in recent years, the need for well-trained IT specialists in the fields of Data Science and Business Analytics is becoming ever greater and the topics are currently highly relevant. — Every Job Recruiter
Many Universities therefore offer part-time education courses to meet the extremely high demand for specialists in data exploration and analysis in the medium term. I studied Data Science at a German university and will give you an insight in the following.
But first let me give an introduction of where I come from and why I chose this way:
I studied business administration in the dual bachelor program and then gained insight into the world of business intelligence through a local Enterprise-Resource-Planing (ERP) manufacturer. They had a small BI department and I was really lucky that my boss at the time offered me a job as a business intelligence consultant, even though I didn’t have great technical skills yet. So I spend a few years with lots of night sessions, some great colleagues and training on the job, to gain more and more experience.
Nevertheless in Europe, and especially in Germany, the assessment of an applicant’s skills is almost exclusively based on academic standards. With that I want to say: If you do not have a degree in a particular subject, you will have a difficult time getting a good job in the market. The academic degree is the entry ticket and counts much more in this country than voluntary additional qualifications or on-the-job training programs.
Therefore my decision was made. To improve my career prospects in the long run, I definitely needed a degree in business information systems. Since I had spent the last 4 years as a data analyst anyway, working primarily using SQL and Power BI, I had created an initial basic understanding. However, I didn’t want to give up my current job for financial reasons either, which is why I decided to do the Master’s part-time.
In 2018, the term Data Scientist was already on everyone’s lips and promised exceptionally good career prospects. So I started looking for opportunities to study for a master’s degree in data science.
Long story short. I found an exciting Data Science Master in Stuttgart (it’s in southern Germany; Hometown of Porsche and Mercedes) and was fortunately able to enroll there successfully.
The Master of Science focuses on both Data Science and Business Analytics to integrate the scientific methods from statistics, computer science, data-based management and decision making.
The curriculum provides students with in-depth study in business intelligence and Big Data technologies and their applications and procedures. The study content includes new, theoretical knowledge about business models as well as the testing of cloud-based tools and infrastructures up to concrete applications, which are developed in workshops with global industry partners.
In the next step I would like to discuss the learning concept consisting of individual modules and their components:
The program is divided into 4 semesters + master’s thesis. A semester consists of three modules, each lasting two months and conducted sequentially. Within the framework of three attendance days (mostly Thursdays, Fridays, Saturdays) per module, new topics and new content are taught. At the end of a module there is another attendance day, on which the certificate of achievement is provided. In workshops, concrete tasks are worked on in the business intelligence and big data labs of the university. In general, the focus of this studies was more on the practical side, which I liked very much, because I could do my own research and work with the data immediately.
Modul 1:
Business Analytics
In this module, selected approaches to business analytics are covered in depth. The module focuses on strategic and organizational aspects of enterprise-wide business intelligence solutions as well as on methodological and technical approaches. One focus of this module is on the extraction, integration and analysis of enterprise-wide distributed, inhomogeneous, structured data.
Used Frameworks:
- Azure Data Factory Pipelines
- SSAS Multidimensional Mode
- Power BI
Applied Statistics
This module focuses on the application of methods from statistics. With their help, large amounts of data from various sources can be responsibly and objectively translated into information and knowledge. In particular, the module provides knowledge of the most important statistical model classes and analysis tools for the practical analysis of complex data, as well as the ability to develop solutions for new problem classes.
Used Frameworks:
- YAML
- Xaringan
- R
Introduction to Data Science
The Data Science module is about fundamentals of Data Science and Big Data. The analysis of large amounts of data in different formats and preferably in real time requires new processes, methods and tools. These different facets of Data Science and Big Data are considered in this module and deepened in practical exercises. I can definitely recommend the Azure Machine Learning Cheatsheet which helped me a lot.
Used Frameworks:
- Azure Machine Learning Studio
- Rapidminer
- Github
Modul 2:
Data Warehouse Workshop
This module deals with the evaluation of different data warehouse architectures and the concrete construction of a data warehouse system. A data mart is built up step by step on the basis of a reference architecture. This is based on a data store. A central component of the module are concepts for modeling multidimensional data structures and their transition into different logical data models. Extraction, transformation and loading processes are used to fill a data mart from an ERP source system.
Used Frameworks:
- SAP Cloud Datawarehouse
- MongoDB
- PostgreSQL
- Dbeaver / PgAdmin
- Google BigQuery
- ER-Model / Logical Model
BI and Big Data Design Workshop
This module deals with the use and methodical design of hybrid BI and Big Data architectures. In particular, the processes and structures for developing and operating BI and Big Data architectures are presented. I learned how Big Data projects are created, designed and managed. Furthermore, how competence centers are managed. A Big Data project will be designed and prototypically implemented. Therefore we used Azure with the corresponding toolset:
Used Frameworks:
- Data Factory v2
- Data Lake
- Hadoop / Hive
- SQL Server
- Azure Data Studio
- SSAS Tabular Mode
Python for Data Science
This module is about getting to know the concepts of current programming languages and their development environments for applications in the data science environment. The module focuses on the basics and differences of the programming languages Python. Students will learn about and apply the data and control structures in Python. A simple problem in Data Science will be solved using the programming language Python. During the time I really fell in love with the Python framework. A summary of Python’s main features is also given in 6 Reasons Why Phyton is Suddenly Super Popular. The stackoverflow post The Incredible Growth of Python shows that Python is currently the most growing programming language. Reasons for this overwhelming growth are elaborated in the post Why is Python Growing so Quickly?.
Used Frameworks:
- Python
- Anaconda
- Jupyter Notebook
- Numpy / Pandas / Matplotlib / Scikit learn / PyCharm / Seaborn / Bokeh / Plotly
Modul 3:
BI and Big Data Architectures
This module is all about architecture options, technologies and systems in the area of Business Intelligence (BI) and Big Data. Students will learn about different options for building a BI and Big Data architecture in order to decide which technology to use in which situation. The module focuses on programming interfaces and data models in the so-called “Apache-Hadoop Ecosystem”, HBase, HIVE and SPARK. We learned about tools for NoSQL databases (MongoDB) and in-memory databases (SAP HANA). Also we used the Google cloud and focused on:
- GCS (Google Cloud Storage, Creation of buckets and loading through various tools)
- Google Big Query
- DataPrep
- DataStudio
- DataProc (Create Hadoop Clusters. Master Node Worker Nodes)
- PySpark (Query big data in a SQL manner)
- Cloud Pub/Sub (Distribute data streams to different endpoints)
- Google Cloud Data Flow
- Google Cloud Data Fusion
Web and Social Media Analytics
In this module, participants deal with the collection, analysis and integration of structured and unstructured data from the Internet. In general, this data comes from websites, in particular from social networks (Facebook, Twitter, Communities, etc). Practical exercises are used to integrate data from the Internet into a data warehouse.
Used Frameworks:
- Beautiful Soup
- Puppetteer
- Twitter API
- Tableau
Machine Learning in Python:
This module is about getting to know and applying algorithms for machine learning as well as for data mining methods in detail. For this purpose, both the methods and the possibilities for parameterizing the methods are presented. Methods and algorithms from different machine learning paradigms are introduced and exemplary applications are presented. In particular, the machine learning process was presented. This includes topics such as normalization and standardization, as well as OneHot or label Encoding. Also, the theoretical treatment of unbalanced datasets was discussed and solutions were shown.
Overall, this was one of the most exciting modules!
Used Frameworks:
- Keras
- Tensorflow
- PySpark
- SciKit
- Docker / Kubernetes
- Deep Learning / Neuronal Networks
Modul 4:
Programming Languages Algorithms and Implementation:
This module is about getting to know and applying algorithms for machine learning and data mining methods in detail. For this purpose, the methods as well as the possibilities for parameterization of the methods are being taught. Procedures and algorithms from different paradigms of machine learning with exemplary applications are presented.
Used Frameworks:
- SQL
- Python
- R
- RapidMiner
New Business Models and Strategies
In this module, we learned to apply the methodology of business model innovation (meta-model, process model, techniques and results) to the insights of Big Data and business analytics solutions, and thus to develop and implement our own business model. In addition, we learned to independently reflect on the results obtained.
Used Frameworks:
- Business Modell Canvas
- Design Thinking
- Power BI
Ethics and Law
This module provides students with in-depth knowledge of business ethics in the context of data analysis. They will acquire the knowledge and skills to examine cause and effect relationships from an ethical perspective and to make decisions in an ethical context. In addition, legal aspects such as national and international data protection standards and laws are taught. You will learn about personal rights and be able to implement measures to comply with these rights.
I enjoyed the master’s program extremely much. The high proportion of practical experience in particular was a key criterion for why I chose it. We didn’t have to write any traditional papers. Instead, we were allowed to work on a practical project with our know-how in each module and present it afterwards.In addition, we usually had to write a project report that was also evaluated.
That was my little description of my dual Data Science Master. I hope you could take away some useful hints. If you are interested in my assessment of career opportunities and development prospects, feel free to write it in the comments.
I am looking forward to your suggestions!
