Project optimization of business processes / Project optimalisatie van bedrijfsprocessen
Supervisors: Rene Bekker,
Objective: Acquiring skills and experience necessary for building decision support systems.
Skill: Use of Visual Basic for Applications (VBA) in combination with Excel.
Goal: Building (and design) of a Decision Support System (DSS) in Excel that:
- is from all perspectives (system design, algorithm, etc.) scientifically correct;
- is useful in practice.
Required time: maximal 160 hours in the 4 weeks of part 3.
Students have to register the time that they have spent on the project.
- week 1: VBA plus making plans and determining contents deliverable(s).
- week 2 and 3: executing plans.
- week 4: making final changes, documentation, implementation.
Monday 3 January 11.00: introductory meeting (P6.47).
Friday 7 January 15.00-17.00, rooms S101 and S103: individual test VBA (Alwin & Rob).
Monday 10 January 13.00-14.30: presentation, discussion and registration of the project plan.
Friday 14 January 17.00: deadline final version of the plan.
Friday 14 January: discussion about the progress in the office of the supervisor.
Friday 14 January 13.00-15.00, room S353: re-exam VBA.
Friday 21 January 13.30-16.30: presentation prototype(s).
Friday 28 January 17.00: deadline deliverables.
Monday 31 January 11.00-13.00: final exam with the students.
During this project students will learn to use Visual basic.
The students will learn the basic skills, such as writing a function and dialog windows, during the first week of the project.
On friday in week 1 each student has to fullfill an individual exam.
The exam consists of writing a simple VBA application that implements a mathematical function and contains a graphical user interface.
A good preparation is required!
Literature: Visual Basic help in Excel, An introduction to VBA in Excel by Robert McDonald, VBA-handleiding by Theo Peek (in Dutch), VU documentation, extra literature (available from the library and supervisors).
You are allowed to use the literature during the exam.
The purpose of the projects is to build a set of Excel function that have the potential to be applied to business situations.
The user is a planner with experience in Excel, who does not find desired solutions with standard software; the focus is the functionality, with a flexible interface, and the tool is not necessarily foolproof.
Due to time restrictions maybe not all of the functionality is implemented at the end of the project.
In that case, student have to specify the missing parts.
The deliverables of each project consists of:
- Project plan;
- Final report;
- Excel functions or add-ins that are public accessible on the internet.
At the end each group sends all files (documentation, programs, etc.) in 1 zip-file to the supervisor.
The project plan consists of the following parts:
- Functions to build;
- Task separation;
- Explanation of the decisions made.
The final report contains the following parts (can exists of multiple documents):
- User manual of the DSS;
- Documentation of the DSS;
- Scientific explanation of the algorithms used in the DSS;
- Report about the verifications of the functions.
In addition, two presentations are obligated, using a beamer, powerpoint and Excel/VBA. The subjects are:
- project plan;
Language of the reports and the DSS: English.
During the project the students are allowed to contact their supervisor, for example to answer questions.
Finally, there is a discussion about the progress and an oral exam with the student.
Criteria and Grading
The following skills play an important role in this project and they are used
- applicability and understanding of the DSS in practice (35%)
- analytical skills: understanding of the theory used for the DSS (20%)
- initiatives and creativity in relationship to the difficulty of the topic and
to the availability of documentation and literature (15%)
- social skills and presentation skills: team working, scientific writing
skills, clearness of documents, understanding during presentations (15%)
- VBA-exam: structure in and clearness of VBA-code (15%)
The first criterion is a group mark, while the other skills are graded
Project 1: Sojourn times in stochastic activity networks
Queueing networks and activity networks found their way in many application areas, including telecommunication systems, production systems, project planning, and health care. In an activity network, a set of several activities or tasks have to be carried out following some precedence relations. In addition to sequential activities, the precedence relations may involve probabilistic branching and parallel branching with `AND' and `OR' type join, see e.g. Choudhury and Houck (1994). In stochastic activity networks, the activity durations are random and there might be a waiting time before an activity starts due to limited resources. Specifically, an individual component might either be a single server of an infinite-server queue (the latter representing a delay).
The aim of this project is to determine the sojourn time distribution of a stochastic activity network. For activities represented by a single-server queue, the activity duration (excluding the queueing component) may be assumed to be exponential. For infinite-server queue, it should be more general. The sojourn time distribution for such a network in general is intractable, but Choudhury and Houck (1994) present an approach based on approximating the activity durations by discrete random variables (Steps 3 and 4.1 of this approach may be modified such that working with LST may be avoided). Hence, the network should be build up using M/M/1 and M/G/infinity queues with precedence relations.
Finally, it would be of interest to compare the results in case of deterministic activity lengths and determine the critical path. This might give an indication of the criticality of each activity on the sojourn time.
- G.L. Choudhury, D.J. Houck (1994). Combined queueing and activity network based modeling of sojourn time distributions in distributed telecommunication systems. ITC 14, 525-534.
Supervisor: Joost Bosman.
Members group 1:
Members group 2:
Project 2: Portfolio selection
In a portfolio selection problem, an investor wants to invest a certain amount of capital (in cash or in holdings in assets). Investors like returns and dislike risks, requiring a diversified portfolio. The aim of this project is to build a tool that supports the investor in selecting the desired portfolio by computing the expected risks and returns for all securities (stocks and bonds) he is considering to buy or sell.
Roughly, the process of selecting a portfolio may be divided into two stages. The first stage involves parameter estimation, or forecasting, of future risks and returns (and correlations). The second stage concerns the computation of risks and returns of an optimal (or efficient) portfolio. The optimal portfolio should be computed for different optimization
objectives and constraints:
- pure return maximization (LP and MIP problems)
- risk minimization (Markowitz problems - QP or QMIP)
- H. Markowitz (1952). Portfolio selection. The Jornal of Finance 7, 77-91.
- Lecture notes on Financial Risk Management: Presentation.
Supervisor: Alwin Haensel.
Members group 3:
Members group 4:
Project 3: Outlier detection in health insurance
Outlier detection refers to the problem of finding pattern in data that do not conform to expected behavior. Some applications areas where outlier detection is used are fraud detection for credit cards, insurance or health care, intrusion detection for cyber-security, and fault detection in safety critical systems, see e.g. Chandola et. al. (2009). The aim of this project is to build a tool that identifies outliers in a database of health care insurance. In addition, the tool should be able to find the records that are most related to potential outliers, that is, the k-nearest neighbors. Such a tool should be designed to assist a health care insurance company in focusing on particular patients that exhibit non-normal claim behavior.
A useful survey on outlier detection is Chandola et. al. (2009). Sections 1 and 2 may provide a useful general introduction to the topic. The simplest techniques focus on anomalies in 1 dimension or use score functions. Other techniques are based on histograms (see Subsection 7.2.1) or k-nearest neighbors (see Section 5).
Finally, more advanced approached based on nearest neighbors are Breuning et. al. (2000) and Papdimitriou et al (2003).
- V. Chandola, A. Banerjee, V. Kumar (2009). Anomaly Detection: A Survey. ACM Computing Surveys 9, 1-71.
- M.M. Breuning, H. Kriegel, R.T. Ng, J. Sander (2000). LOF: Identifying density-based local outliers. Proceedings of ACM SIGMOD, Texas.
- S. Papadimitriou, H. Kitagawa, P.B. Gibbons, C. Faloutsos (2003). LOCI: Fast outlier detection using local correlation integral. Proceedings of the 19th ICDE.
Supervisor: Rob Konijn.
Members group 5:
Members group 6: