Michael Vössing

Motivation

In the gastronomy sector, wages account for about one-third of all costs. Accordingly, restaurateurs try to optimize the utilization of their employees by creating appropriate work schedules. A variety of commercial tools are available to create these work schedules. However, many tools require restaurateurs to manually estimate how many employees are needed at any given time. Estimating the "fluctuating'' personnel requirement that depends on a variety of factors (i.e., weather, events) is difficult. As a result, many restaurateurs struggle with either too high labor costs (i.e., over-staffing) or do not realize their revenue potential (i.e., under-staffing). Nesto has developed software for intelligent and automatic personnel planning. Their product provides a simple, intuitive, and efficient solution for the creation of work schedules to help restaurateurs increase their return on sales. One of the unique features of their software is the ability to forecast the revenue of restaurants to address over-staffing and under-staffing proactively. 

Research Goal

As an approximation of a restaurant's personnel requirements, machine learning techniques can be used to predict its hourly revenue based on historical sales data as well as supplementary data. The primary objective of this thesis is to improve an existing forecasting model by compiling, implementing, and evaluating different state-of-the-art techniques for (a) the incorporation of supplementary data into forecasting models, and (b) the explanation of the impact of individual features on the restaurant's revenue. You will...

• apply your theoretical knowledge in a practical environment.
• become an expert on the subject of time series forecasting.
• implement and evaluate forecasting models. 
• explain which features influence the revenue of a restaurant.

If you are interested, check the attached PDF document and send your CV, transcript of records, and a brief letter of motivation to michael.voessing@kit.edu.