Forecasting the number of immigrants on the maritime line and patrol boat routing

This study focuses on the problem of estimating the number of imigrants on Turkey's maritime line and patrol boat routing for the first time for a national security project. Transporting immigrants by sea is one of Turkey's biggest concerns, among other countries in the world. The cost of using various types of vehicles to catch undocumented immigrants can be extremely high. It should be optimised with the concentration of immigrants in the relevant regions in mind, both to minimize the spending problem and to ensure an accurate security screening. In order to achieve this goal, estimating the number of immigrants along the maritime line route in the first stage is seen as a significant problem and possibilistic and non-probabilistic models are studied. Since the data for 2020 and 2021 have not yet been accessed due to COVID-19, seven distinguished forecasting methods are applied from simple to complex to obtain estimates of the 2019 data set and to compare actual observations using dataset between 2016 and 2019 obtained from the Coast Guard Command within the Ministry of Interior of the Republic of Turkey. Then, the Forecast Combination Approach with Meta Fuzzy Functions, which combines all methods, is proposed. According to the final assessments, the proposed approach provides more accurate forecast results for the expected number of migrants on Turkey's maritime line, which are used for patrol planning. In the second step, we identify the hotspots by clustering the coordinates of the observations from 2016-2019 and solve a mixed vehicle routing problem to visit these hotspots. We then address the situation where the Command's resources are limited and propose an approach to prioritize hotspots based on their criticality. We use the estimation results of the proposed method obtained in the first step as the criticality weight of the hotspots. We integrate these weights into a steering formulation that takes into account the detection power and the total distance traveled simultaneously. With a bioobjective approach, we produce alternative efficient solutions at different distance and detection power levels that can be taken into account by the decision makers of the problem. This is the first study in Turkey that aims to create efficient maritime patrol routes that will reduce costs and increase national security. We present a comprehensive methodology for this problem integrating clustering, prediction and multi-objective optimization techniques.