This thesis involves the development and evaluation of a software approach to detect annomalies in the walking pattern of cows using a wearable motion sensor. As part of this thesis, a study should be conducted in a stable in Oberschleissheim to collect cow data (we will do this together). Afterwards, we will analyze the cow's data and develop an approach to detect anomalies in a cow's walking pattern. Experience in machine learning is a plus but not required.

Software metrics have been proven to be useful in indicating problems in development projects, but they are limited to analyzing information captured in software artifacts. By applying metrics to communication artifacts, we want to make the development process more insightful, validate the software metrics by combining them with results derived from aspects of team communication, and make it possible to recognize problems from a very early project stage.

The team allocation for software projects is often done solely based on the subjective judgment of the project manager. The objective of this thesis is to create a hybrid system called TEASE (TEam Allocation for Software Engineering). It makes use of an algorithm while allowing user interaction to create optimal teams from a pool of persons. The project manager can adapt the generated team allocation through an intuitive user interface based on his own knowledge and experience. The focus of this thesis is the user interface of TEASE.

The problem of team allocation is a complex task for humans to manage. We address this problem by exploring possible algorithmic solutions and implementing some of them in an interactive decision-support system. We collected a range of algorithmic approaches, introduced a convenient cat- egorization of them, and implemented one of the most promising ones into an actual software solution. We further held two distinct experiments with the implemented system. The results of these experiments show that the sys- tem finds an optimal team allocation with regards to the specified allocation objective and satisfies all specified constraints.

User feedback is difficult to use as a basis for implementation. For this purpose, we introduce FeedViz, a feedback visualisation component which analyses, structures and prioritises user feedback and illustrates the results to the development team. The aim is to simplify the decision-making process as well as to capture information related to feedback which is important for rationale management.

Studying different methods to visually detect the damages on metallic surfaces, from simple background subtraction to supervised and unsupervised machine learning approaches, and performing an experiment to compare the results.