Seminar for
DECISION MAKING – THEORY, TECHNOLOGY AND PRACTICE

 

PROGRAM

---

Predavanja možete pratiti i online putem MITEAM stranice Seminara Odlučivanje - teorija, tehnologija, praksa:
https://miteam.mi.sanu.ac.rs/asset/sEL32w8mjmruyeEqW

---

Plan rada Seminara Odlučivanje - teorija, tehnologija, praksa za NOVEMBAR 2025.

---

Četvrtak, 13.11.2025. u 13:00, Pariske Komune bb, Niš i Online
Nataša Milosavljević, University of Belgrade, Faculty of Agriculture
DISTANCE BASED ON MATHEMATICAL LOGIC IN THE CONTEXT OF MISSING DATA: APPLICATION ON PUBLICLY ACCESSIBLE UCI DATABASES WITH THE SUPPORT OF METAHEURISTIC ALGORITHMS
Missing data represent one of the most common challenges in real-world data analysis and significantly affect the reliability and accuracy of predictive and classification models. This lecture will present a new approach to defining the gap between data containing incomplete information, based on mathematical logic, taking into account the degree and type of missing values. The proposed distance was tested on several UCI (UCI Machine Learning Repository) datasets containing partially or completely missing data. As a part of the research, various metaheuristic algorithms were applied in order to optimize the classification and clustering performance with the newly defined distance. Comparison with classical methods for processing missing data (such as mean imputation, kNN-imputation, etc.) shows that the new approach provides greater stability and accuracy in scenarios with a high percentage of missing values. Special attention is paid to sensitivity analysis in relation to the percentage of missing data, as well as to determination of algorithm’s parameters through adaptive mechanisms.
The presented results were obtained in collaboration with M. Rašković and T. Davidović.
Joint Seminar meeting with Computer Science and Applied Mathematics and Artificial Intelligence.

---

Četvrtak, 20.11.2025. u 13:00, Online
Dragan Urošević, Mathematical Institute of the Serbian Academy of Sciences and Arts
THE GENERAL VARIABLE NEIGHBORHOOD SEARCH FOR THE COLORED TRAVELING SALESMAN PROBLEM
Given a set $U$ containing n customers, labeled from $1$ to $n$, and a set $T$ of $m$ traveling salesmen, labeled from $1$ to $m$ $(1<m<n)$. The customers are divided into $m+1$ disjoint sets: $C_1$, $C_2$, ..., $C_{\mathrm{m}}$, and $S$. All salesmen start from a depot (labeled as $0$) and must return to the depot after visiting their assigned customers. All customers in set $C_i$ must be visited by the salesman labeled $i$ (for $i=\mathrm{1,2,...,}m$), while customers in set $S$ can be visited by any salesman (but each must be visited by exactly one salesman). The Colored Traveling Salesman Problem (CTSP) consists of determining m routesfor the m salesmen (in accordance with the given constraints) so that the total length of all m routes is minimized. A Variable Neighborhood Descent (VND) method using a set of neighborhoods was developed and incorporated into a General Variable Neighborhood Search (GVNS). The proposed method was tested on standard benchmark instances grouped into three categories. The initial results show that the method is efficient and produces very good solutions.

---

Četvrtak, 27.11.2025. u 13:00, Online
Željko Stević, University of East Sarajevo, Faculty of Transport and Traffic Engineering
MANAGEMENT OF LAST-MILE DELIVERY THROUGH THE SELECTION OF LOCATIONS FOR PARCEL LOCKERS USING MCDM MODEL
When observing e-commerce, it can be concluded that parcel lockers are an innovative solution to the everyday challenges faced in logistics. The fast pace of life and traffic congestion make parcel delivery difficult and expensive, while parcel lockers offer a more efficient way to receive and send packages 24/7. Key advantages include reduced delivery costs, environmental benefits (less ${\mathrm{CO}}_2$ emissions), security of packages and greater availability to users. Parcel lockers are making access easier, while couriers can optimize their delivery routes. In this research, using the FUCOM (Full Consistency Method) and MARCOS (Measurement Alternatives and Ranking According to the Compromise Solution) methods, a model for selecting optimal locations for installing parcel lockers of the X-Express company for the Vitez distribution center was developed. The first step is data collection, while multi-criteria decision-making methods enable detailed analysis and ranking of potential locations. The introduction of parcel lockers contributes to the modernization of logistics, reduces administrative tasks and supports environmental sustainability.

---

---