Designing Efficient Stratified Mean-Per-Unit Sampling Applications in Accounting and Auditing

Despite technological advances in accounting systems and audit techniques, sampling remains a commonly used audit tool. For critical estimation applications involving low error rate populations, stratified mean-per-unit sampling (SMPU) has the unique advantage of producing trustworthy confidence intervals. However, SMPU is less efficient than other classical sampling techniques because it requires a larger sample size to achieve comparable precision. To address this weakness, we investigated how SMPU efficiency can be improved via three key design choices: (a) stratum boundary selection method, (b) number of sampling strata, and (c) minimum stratum sample size. Our tests disclosed that SMPU efficiency varies significantly with stratum boundary selection method. An iterative search-based method yielded the best efficiency, followed by the Dalenius–Hodges and Equal-Value-Per-Stratum methods. We also found that variations in Dalenius–Hodges implementation procedures yielded meaningful differences in efficiency. Regardless of boundary selection method, increasing the number of sampling strata beyond levels recommended in the professional literature yielded significant improvements in SMPU efficiency. Although a minor factor, smaller values of minimum stratum sample size were found to yield better SMPU efficiency. Based on these findings, suggestions for improving SMPU efficiency are provided. We also present the first known equations for planning the number of sampling strata given various application-specific parameters.

Methodology for Optimal Placement of Containers in Trains in Case of Cargo Operations Along the Route

A new approach to organisation of container block trains is considered based on the principles of passenger traffic. The technology assumes container train’s traffic subject to the timetable with sale of cargo space in the train. The train is made up at the departure station and follows the established route with stops at intermediate container terminals or stations, where a container for which this station is designated as destination is removed and a new container is placed on the vacated place to be delivered to subsequent points of the route.The objective of this study is to develop a methodology for optimal placement of containers in a block train intended for en route cargo handling operations. The technique involves an iterative search for such an order of placement of packages so that containers assigned to each intermediate point are as close to each other as possible. The technique is an authors’ algorithm based on combinatorial optimisation methods.The implementation of the proposed algorithm makes it possible to reduce the excessive mileage of handlers and loaders at intermediate points and, consequently, to increase speed of cargo operations when rearranging containers, as well as to reduce operating costs of using the loading facilities of the container terminal.The proposed mathematical algorithm as compared to exhaustive search allows significantly reducing the number of iterations in search for a solution and can be implemented as software.

General Purpose Optimization Library (GPOL): A Flexible and Efficient Multi-Purpose Optimization Library in Python

Several interesting libraries for optimization have been proposed. Some focus on individual optimization algorithms, or limited sets of them, and others focus on limited sets of problems. Frequently, the implementation of one of them does not precisely follow the formal definition, and they are difficult to personalize and compare. This makes it difficult to perform comparative studies and propose novel approaches. In this paper, we propose to solve these issues with the General Purpose Optimization Library (GPOL): a flexible and efficient multipurpose optimization library that covers a wide range of stochastic iterative search algorithms, through which flexible and modular implementation can allow for solving many different problem types from the fields of continuous and combinatorial optimization and supervised machine learning problem solving. Moreover, the library supports full-batch and mini-batch learning and allows carrying out computations on a CPU or GPU. The package is distributed under an MIT license. Source code, installation instructions, demos and tutorials are publicly available in our code hosting platform (the reference is provided in the Introduction).

Bayesian Optimization Approaches for Identifying the Best Genotype from a Candidate Population

Bayesian optimization is incorporated into genomic prediction to identify the best genotype from a candidate population. Several expected improvement (EI) criteria are proposed for the Bayesian optimization. The iterative search process of the optimization consists of two main steps. First, a genomic BLUP (GBLUP) prediction model is constructed using the phenotype and genotype data of a training set. Second, an EI criterion, estimated from the resulting GBLUP model, is employed to select the individuals that are phenotyped and added to the current training set to update the GBLUP model until the sequential observed EI values are less than a stopping tolerance. Three real datasets are analyzed to illustrate the proposed approach. Furthermore, a detailed simulation study is conducted to compare the performance of the EI criteria. The simulation results show that one augmented version derived from the distribution of predicted genotypic values is able to identify the best genotype from a large candidate population with an economical training set, and it can therefore be recommended for practical use. Supplementary materials accompanying this paper appear on-line.

Optimizing large scale bin packing problem with hybrid harmony search algorithm

Bin packing problem (BPP) is a combinatorial optimization problem with a wide range of applications in fields such as financial budgeting, load balancing, project management, supply chain management. Harmony search algorithm (HSA) is widely used for various real-world and engineering problems due to its simplicity and efficient problem solving capability. Literature shows that basic HSA needs improvement in search capability as the performance of the algorithm degrades with increase in the problem complexity. This paper presents HSA with improved exploration and exploitation capability coupled with local iterative search based on random swap operator for solving BPP. The study uses the despotism based approach presented by Yadav et al. (2012) [Yadav P., Kumar R., Panda S.K., Chang, C. S. (2012). An intelligent tuned harmony search algorithm for optimisation. Information Sciences, 196, 47-72.] to divide Harmony memory (HM) into two categories which helps to maintain balance between exploration and exploitation. Secondly, local iterative search explores multiple neighborhoods by exponentially swapping components of solution vectors. A problem specific HM representation, HM re-initialization strategy and two adaptive PAR strategies are tested. The performance of proposed HSA is evaluated on 180 benchmark instances which consists of 100, 200 and 500 objects. Evaluation metrics such as best, mean, success rate, acceleration rate and improvement measures are used to compare HSA variations. The performance of the HSA with iterative local search outperforms other two variations of HSA.

Modeling of reasoning when searching for objects in images

Visual patterns, for example, handwritten letters or objects of aerospace observations, are highly variable. The high variety and large volume of unstructured information lead to the need for complex and resource-intensive calculations. Unfortunately, image analysis approaches based on the domain ontology do not specify any method for automatic selection of criteria (features) and decision-making rules. Insufficient structuredness of cases and a large variability of object images lead to a rapid growth of the case base, which significantly reduces the performance of the decision support system. The article proposes an approach to the structural analysis of images, which consists in sequential refinement of objects' features and weakening of interpretation rules during an iterative search of facts using the ontology of images represented as attributed graphs of relationships between elements of objects. The algorithm of reasoning on graphic information consists in the sequence of task (functional) actions necessary for processing and analyzing the image in accordance with the task, the actions of the system to prepare conditions for their implementation, as well as to organize and manage the reasoning process.