Learning objectives

This course teaches students how to apply machine learning concepts to develop predictive models that form the basis of many innovative service offerings and business models today. Using a selected use case each term, students learn the foundations of selected algorithms and development frameworks and apply them to build a functioning prototype of an analytics-based service. Students will become proficient in writing code in Python to implement a data science use case over the course period.

Description

Data-driven services have become a key differentiator for many companies. Their development is based on the increasing availability of structured and unstructured data and their analysis through methods from data science and machine learning. Examples comprise highly innovative service offerings based on technologies such as natural language processing, computer vision or reinforcement learning.

Using a selected use case, this lecture will teach students how to develop analytics-based services in an applied setting. We teach the theoretical foundations of selected machine learning algorithms (e.g., convolutional neural networks) and development concepts (e.g., developing modeling, training, inference pipelines) and teach how to apply these concepts to build a functioning prototype of an analytics-based service (e.g., inference running on a device). During the course, students will work in small groups to apply the learned concepts in the programming language Python using packages such as Keras, Tensorflow or Scikit-Learn.

Recommendations

The course is aimed at students in the Master's program with basic knowledge in statistics and applied programming in Python. Knowledge from the lecture Artificial Intelligence in Service Systems may be beneficial.

Additional information

Due to the practical group sessions in the course, the number of participants is limited. Further information on the application process can be found on the course website (https://dsi.iism.kit.edu/64_411.php).

Please apply via the WiWi Portal until April 17, 2020: https://portal.wiwi.kit.edu/ys/3539 

Recommendations:

The lecture is addresed to students with interests and basic knowledge in the topics of Operations Research, decritptive and inductive statistics.

• The Geometry of Multivariate Statistics, Wickens, T. D., Psychology Press, 2014.
• Data Mining: Concepts and Techniques, Han, J., Pei, J., Kamber, M., Elsevier, 2011.
• Data Mining and Analysis, Zaki, M. J., Meira Jr, W., Meira, W., Cambridge University Press, 2014.
• An Introduction to Statistical Learning with Applications in R, James, G. et al., Springer, 2013.
• Forecasting – Principles and Practice, Hyndman, R. J., Athanasopoulos, G., OTexts, 2018.
• Fundamentals of Predictive Text Mining, Weiss S. M. et al., Springer, 2015.
  

Paper:

• How Big Data can make Big Impact: Findings from a systematic review and a longitudinal case study. International Journal of Production Economics, 2015.
• Business Intelligence and Analytics: from Big Data to Big Impact, Chen, H. et al., MIS quarterly, 2012.
• Building Watson – An Overview of the DeepQA Project, Ferrucci, D. et al., AI Magazine, 2010.
  

Further readings will be provided in the lecture.

Today's service-oriented companies are starting to optimize the way services are planned, operated, and personalized by analyzing vast amounts of data from customers, IT-systems, or sensors. As the statistical learning and business optimization world continues to progress, skills and expertise in advanced data analytics and data and fact-based optimization become vital for companies to be competitive. In this lecture, relevant methods and tools will be considered as a package, with a strong focus on their inter-relations. Students will learn to analyze and structure large amounts of potentially incomplete and unreliable data, to apply multivariate statistics to filter data and to extract key features, to predict future behavior and system dynamics, and finally to formulate data and fact-based service planning and decision models.

More specifically, the lessons of this lecture will include:

• Co-Creation of Value Across Enterprises
• Instrumentation, Measurement, Monitoring of Service Systems
• Descriptive, predictive, and prescriptive Analytics
• Usage Characteristics and Customer Dynamics
• Big Data, Dimensionality Reduction, and Real-Time Analytics
• System Models and What-If-Analysis
• Robust Mechanisms for Service Management
• Industry Applications of Service Analytics

Tutorials
Students will conduct lecture accompanying, guided exercises throughout the semester.

The total workload for this course is approximately 135.0 hours. For further information see German version.

Participants are able to structure large sets of available data and to use that data for planning, operation, personalization of complex services, in particular for IT services. They learn a step-by-step approach starting with analyzing possibly incomplete data, techniques of multivariate statistics to filter data and to extract data features, forecast techniques, and robust planning and control procedures for enterprise decision support.

The assessment consists of a written exam (60 min) (according to §4(2), 1 of the examination regulation).