Pr. Wartenberg (Hochschule Hannover) - Dataiku as a Versatile Platform for BI & Beyond

MW

Name:
Maylin Wartenberg

Title:
Professor. Doctor.

Country:
Germany

Organization:
University of Applied Sciences and Arts Hannover (Hochschule Hannover)

Description:
With around 10.000 students the Hochschule Hannover is the second largest university in Hanover, the capital of the federal state of Lower Saxony. Institutionalized in 1971 from various educational precursors – the oldest dating back to the year 1791 – the Hochschule Hannover offers a particularly wide range of subjects in five faculties.

The degree course 'Business Information Systems' is oriented towards business information systems as an independent discipline. The experience that business information systems specialists must have both thorough knowledge of business administration and a comprehensive basic knowledge of computer science is appropriately implemented through the contents of the degree course. One of the specializations to choose from is 'Business Intelligence'.

Awards Categories:

• Excellence in Teaching
  
  

Challenge:

I teach a class in a Bachelor's degree on advanced topics in Business Intelligence. Students focus on ‘Business Information Systems’ and can choose ‘Business Intelligence’ as a specialization, with two courses: ‘Data Warehouse’ and ‘Business Intelligence’.

In my course, Advanced topics in BI, I want to give the students a good overview about possible further topics in the field of data science, not only theoretically, but with lots of practical experience. The topics include different aspects of data science like data preparation, data visualization, and data analysis, as well as topics such as data governance, collaboration, code languages, machine learning, and even deep learning. Therefore I needed a software which facilitates many different use cases based on a broad variety and the interests of the students, and is easy to work with without prior experience.

In addition to that, I teach another class in a Master’s degree called ‘Digital Transformation’. This is a consecutive Master based on a Bachelor’s degree in Business Informatics. Some students have a background in business, others in IT. The topic I teach is an introduction to artificial intelligence. Some of the students already have experience in machine learning topics and are able to program in Python, but some do not have any experience regarding AI. Therefore it was difficult to work out hands-on exercises with such differences in prior experience.

Solution:

I have been using Dataiku DSS for over 3 years in teaching for both classes, and it works very well. It only takes a few basic tutorials to get to know the software and to be able to work with it. The students can even work on complex machine learning tasks within one semester. They have the ability to use the integrated algorithms, or code their own. Dataiku DSS offers such a great variety of topics in tutorials, documentations, and articles so that the students are able to get to know many different aspects of working with data. Each year I try to work with the newest version and constantly explore new additions to the software.

Impact:

I would like to share this year's experience in my class ‘Advanced Topics of BI’ as an example of the possibilities. The students work in small groups and each group gets a special topic. They have to present on the topic in general, and then create a small hands-on workshop for other students in the class.

This year, the topics were:

1. Data Visualization & Storytelling, especially Waterfall, Treemap, and Sunburst Charts
2. Geospatial Analytics, especially MapCharts using Reverse Geocoding/Admin maps
3. Graph Analytics, especially Social Network Analysis
4. Exploratory Data Analysis using Interactive Statistics
5. Data Governance
6. Connectivity/Data Sources, especially SQL Data Tables and SQL Recipes
7. ETL using recipes based on Code, especially Python recipes
8. Code - Notebooks, especially Python Notebooks
9. Webapps, using Dash
10. Deep Learning - Different Libraries, and especially Keras
11. Natural Language Processing - Sentiment Analysis

It really is a wide variety of topics that can be addressed within one software framework. The students create their own scenarios and create or find their own data as the setting for their workshop. Some integrate pictures in their project overviews as you can see here:

Some use the Wiki for describing the tasks in the workshop:

Some use data that already creates an interest in the topic for other students like the Social Network Analysis on the Marvel Universe or Game of Thrones Data:

The feedback of the students regarding the software is always very good. They can be very creative and get to know different topics in data science. The added hands-on experience makes the presentation more interesting and increases the learning experience.