Scribd
Upload
5 views

Weekly-Notes

The document presents a decomposition-based metaheuristic approach for solving the multi-cover routing problem in rapid needs assessment following disasters. It outlines a three-stage algorithm involving site selection, team assignment, and routing, which outperforms existing benchmarks in both solution quality and computational efficiency. The project also includes phases for proposal, implementation, analysis, and presentation, emphasizing a strong understanding of the mathematical model and its real-world applications.

Uploaded by

dastech1998
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
5 views

Weekly-Notes

The document presents a decomposition-based metaheuristic approach for solving the multi-cover routing problem in rapid needs assessment following disasters. It outlines a three-stage algorithm involving site selection, team assignment, and routing, which outperforms existing benchmarks in both solution quality and computational efficiency. The project also includes phases for proposal, implementation, analysis, and presentation, emphasizing a strong understanding of the mathematical model and its real-world applications.

Uploaded by

dastech1998
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 5

A decomposition-based

metaheuristic approach for


solving rapid needs assessment
routing problem
Yurtsev Mıhçıo˘glu *, Erinç Albey Department of Industrial Engineering,
Ozyegin University, 34794 Istanbul, Turkey

Project Aim:
 Solve multi-cover routing problem in the case of rapid need
assessment.
 Rapid need assessment is the process of evaluating the impact of a
disaster in a given site(s) to determine the needs of differente
community groups.

 The paper develops a solution approach that finds the best


assessments in a short period of time.

 The proposed algorithm is a three-stage decomposition:

- Stage 1: Site selection


- Stage 2: Team assignment
- Stage 3: Routing

 The performance of the algorithm is evaluated based on benchmark


algorithms under different instances
 There are 2 stages of needs assessment:
- Rapid needs assessment stage( should start immediately after a
disaster occurs and should lead to a design of a plan based on the
needs of the affected people within three days, at the latest)
- Detailed needs assessment stage
 Site visit teams: Experienced generalists of humanitarian
organizations, visit the affected sites to make observations (e.g., look,
smell, touch) and talk to people (e. g., interviews).
 There are multiple sites affected by the disaster thus, impossible to
apply rapid needs assessment on all of them at the same time
 Humanitarians identify community groups based on some relevant
factors, such as
- Geography,
- Density
- Severity of the impact (may lead to difference in survival needs)
 Sites that carry the specified community groups to be visited are
selected.
 Visiting the same type of community group several times may be
desired to increase the reliability of the evaluations
 In order to solve such a problem, a multi-cover routing problem
(MCRP) assuming a pre-specified coverage target, which represents the
minimum number of site visits for each community group to achieve a
reliable observation for each group, is defined in.
 Steps to solve MCRP by previous literatures:
- Site selection(based on the criteria cited above)
- Gather information and inform stakeholders to enable all
humanitarian actors from the outset.
- The former step requires communication among all teams which
must be in Person (to avoid telecommunication issues)
 Following that, the paper assumes the following:
Remote communication is not possible, and all teams must return to
the coordination center after visiting the sites that they are
assigned to.
 Any solution approach should indulge the three sub-problems:
 Selection of the sites in a way that pre-specified coverage targets for
each community group is met
 Assignment of the selected sites to assessment teams,
 Providing a route to each team that starts and ends at the coordination
center.

 The results recorded indicate that the proposed algorithm can achieve
high-quality solutions expeditiously, providing better results, on the
average, when compared to the best-known solution approaches in the
literature.

 Therefore, we developed a decomposition based tabu search


algorithm, which uses a smart initial solution generation methodology
and a near-exact solution method in solving the third phase (i.e the
third sub-problem) of the problem and utilize this solution to assess the
neighborhood moves as well.
 The contribution of the proposed solution approach is twofold.
 First, provided decomposition based approach enables a modular
structure, in which each sub-problem can be addressed
independently, using the desired solution approach designed to
solve that specific sub-problem. To demonstrate this capability, we
utilize state-of-the-art algorithm, Concorde (Concorde (2015)),
designed to solve traveling salesman problem (TSP).
 Second, the performance of the proposed approach is superior to
that of state-of-the-art algorithms presented in the literature. We
compare our approach to two benchmarks. The first benchmark
provides better quality solutions in longer time, whereas the second
benchmark is faster with worse solution quality.

Results reveal that our algorithm outperforms both benchmarks in terms of


solution quality. Our algorithm runs dramatically faster than the approach
providing the best quality solutions (∼ 10 minutes vs. two hours) without any
deterioration in the attained solution quality.

While providing more than tenfold increase in running time, the solution
quality is improved approximately 1%, which is also found to be statistically
significant (see Section 5.4).

Considering the time criticality of the needs assessment problem, we believe


that this improvement in solution time is quite crucial.

When we compare our results with the faster benchmark, it is seen that our
algorithm performs statistically significantly better (2% improvement on the
average) than the benchmark for the cost of only four minutes of increase in
the run time. All in all, we propose a significantly faster, and statistically
significantly better algorithm with a modular structure that allows ease of
adoption for possible followup research.
PROJECT REQUIREMENTS
The final project will be completed in two phases. Your ability to analyze,
implement, and communicate a research problem effectively will be
evaluated.

Phase 1: Project Proposal (Current Phase) ✅

Declare whether you will work individually or in a group of two. ✅

 Submit the names of the group members (if applicable). ✅

 Select and submit two research papers published by faculty


members of Industrial Engineering Department of Özyeğin University. ✅

 The selected papers can be authored by your advisor or any other


faculty member. ✅

 Ensure that you choose a problem that you can model and solve using
Python. ✅

Phase 2: Implementation & Analysis

1. Understanding the Problem ⏳

o Conduct an in-depth analysis of the problem discussed in the


selected papers.

o Clearly understand the mathematical model, including its


constraints, objectives, and decision variables.

2. Coding the Model

o Implement the mathematical model in Python.

o Solve the model for small-sized instances and analyze the


results.

3. Model Extensions

o Identify possible modifications or enhancements to make the


model more complex and interesting.

o Discuss how these changes impact the


problem's computational complexity and practical
applicability.
4. Presentation & Defense

o Prepare a professional presentation summarizing your


project.

o Be ready to answer questions on any aspect of your work,


including problem formulation, methodology, implementation,
and results.

Key Expectations:

A strong understanding of the mathematical model and its real-world


application.
A well-structured Python implementation with valid solutions for small
instances.
Thoughtful discussion on model improvements and complexity
adjustments.
A clear and confident presentation demonstrating mastery of the
subject.

You might also like