https://andisheyeamari.irstat.ir/ Andishe_ye Amari en daily 1 Wed, 19 Feb 2025 00:00:00 +0330 Wed, 19 Feb 2025 00:00:00 +0330 https://andisheyeamari.irstat.ir/article_733257.html In industrial environments, achieving high-quality production requires continuous monitoring of processes to reduce waste and increase productivity. Statistical quality control serves as an effective tool for process evaluation, and the control chart, as one of its main tools, plays a significant role in identifying and monitoring process changes. The objective of this study is to evaluate the performance of a hybrid control chart obtained by combining Double Moving Average and Exponentially Weighted Moving Average control charts. To assess the performance of control charts, the run length is first calculated using Monte Carlo simulation. Then, the new metrics of Median Run Length and Standard Deviation of Run Length, alongside the common metric of Average Run Length, are obtained for them, and the capability of control charts in detecting small process shifts based on these new metrics has been evaluated. The results indicate that hybrid control charts are more sensitive than those presented in previous literature in detecting small shifts and represent a more effective tool for enhancing quality and continuous improvement of production processes. https://andisheyeamari.irstat.ir/article_732975.html The rapid expansion of mobile communication technologies has led to the continuous generation of large volumes of location data by mobile network users. When used properly and legally, these data provide a valuable source for producing official statistics in areas such as migration, tourism, crisis mobility, transportation, and crisis management. The present paper is designed to explore the potentials and limitations of employing mobile positioning data and to outline a methodological framework for producing statistics based on these data within the national statistical system. Given the lack of Persian references in this field, this study seeks to provide a practical framework for the use of mobile phone data in the production of official statistics through a comprehensive review of international literature and the analysis of a real dataset. Within this framework, various types of mobile phone data and their applications in statistical production are introduced, and the legal, ethical, and technical requirements for their use are examined. Furthermore, the methodology for extracting and processing mobile positioning data is described, and a case study based on a dataset related to urban mobility patterns is presented. The main contribution of this research lies in documenting practical experiences and transferring technical knowledge to support the localization and application of such data within the national statistical system.The case study showed that crises significantly reduce users’ mobility, with fewer individuals moving and travel concentrated in specific areas. https://andisheyeamari.irstat.ir/article_733260.html Traditional statistical methods have faced serious challenges due to the expansion of spatial data with complex spatiotemporal structure. These data require specialized methods due to spatial autocorrelation, variance heterogeneity, and complex geographical dependencies. In this study, support vector regression is introduced as a novel approach for analyzing and modeling the complex spatial structure of geostatistical data related to soil calcium and magnesium contents. This analysis is performed based on different geographical coordinates (east–west and north–south), at two depths of 0–20 cm and 20–40 cm, and across three distinct geographical regions. The support vector regression method, with its capability to model complex nonlinear relationships while preserving the spatial structure of the data, allows for more accurate and realistic prediction of the nutrient elements' distribution in soil. This approach, utilizing kernel functions, enables the analysis of high-dimensional feature spaces and structural complexities, proving its effectiveness against noise and outliers. To accurately measure the efficiency of support vector regression, its performance is compared against ridge regression. https://andisheyeamari.irstat.ir/article_734247.html The principle of insufficient reason, or indifference principle, is not an axiom of probability but the basis for the definition of classical (logical) probability. Bayesian statisticians also follow the principle of using uniform distributions when there is no a priori information about the parameter. Is this principle acceptable to statistical philosophers?. Several paradoxes have been proposed to refute this principle. Although these paradoxes have not been resolved, the principle is still the basis of the most important definition of probability, classical probability. In this article, this principle, its origin, and paradoxes are evaluated. https://andisheyeamari.irstat.ir/article_735098.html This study was conducted with the aim of qualitatively examining and comparing the performance of several AI-based tools in teaching the concepts of variance and standard deviation, using a qualitative approach and case study method. The statistical population consisted of AI-based tools applicable in statistics education. Sampling was purposeful, based on accessibility and the ability to generate statistical content, and five tools—Gemini, ChatGPT, MathGPT, MagicSchool, and DeepSeek—were selected. Data were collected using four input prompts, and to evaluate the AI responses to these prompts, criteria such as clarity, engagement, structure, interactivity, real-world relevance, and appropriateness for students were employed. Data analysis was performed through coding with the assistance of MAXQDA software, and the reliability of the analyses was confirmed with a kappa coefficient of 0.8. The findings indicated that AI tools can enhance statistics learning by providing engaging and real-world-relevant content, though they have limitations such as lack of creativity or insufficient accuracy. Overall, in this domain, the tools exhibited varying performance. Gemini and DeepSeek performed relatively better in delivering precise and coherent lesson plans, ChatGPT in designing analytical questions and attractive visual images, MagicSchool in facilitating group activities and addressing common errors, and MathGPT in providing comprehensible content related to the concepts of variance and standard deviation. Combining these tools with teacher supervision—which is essential for maintaining academic integrity—can strengthen statistical literacy and critical thinking, preparing students to face the challenges of a data-driven world. https://andisheyeamari.irstat.ir/article_736944.html Emergency response time in traffic accidents is one of the key determinants of the quality of pre-hospital emergency medical services. Predicting this time enables decision-makers to improve the efficiency of emergency services and reduce response time. The data related to emergency response time are a type of time-to-event data, whose main characteristic is their dependency on duration. To account for this feature, baseline hazard models are commonly used; however, the performance of these models may be limited due to their underlying assumptions. In contrast, machine learning models serve as an alternative approach for modeling emergency response time. Their main advantage is that they are not constrained by the restrictive assumptions of baseline hazard models and can capture nonlinear and interactive relationships among variables. In this study, a survival-based neural network model was employed to simultaneously estimate the survival function and predict emergency response time. By utilizing the concept of multi-task learning, this model can account for duration dependence while predicting response time through the concurrent estimation of the survival function. Using 28,505 traffic accident reports recorded by the Mashhad Emergency Medical Services, the performance of the proposed model was evaluated and compared with statistical models and other machine learning methods. The results indicate that the proposed model can serve as an effective alternative to traditional approaches for predicting emergency response time. https://andisheyeamari.irstat.ir/article_736945.html Risk management is an important process for making investment decisions, which includes analyzing the risk in an investment and deciding whether or not to accept that risk in light of the expected returns for the investment. There are various parametric, non-parametric and semi-parametric methods for measuring and analyzing risk in financial markets, but the two main and widely used methods are value at risk and expected shortfall. In this article, a method for estimatingthe expected shortfall is presented, in which different value-at-risk estimation methods and ARMA − GARCH time series models are used to predict future fluctuations. Finally, this method is tested on real data and its superiority over the parametric method is shown by the feedback test. https://andisheyeamari.irstat.ir/article_737138.html Modeling nonlinear data, due to the presence of complex relationships and latent structures, requires approaches that go beyond classical linear models. In this study, the Bayesian neural network framework is employed to evaluate the performance of different Bayesian inference methods for nonlinear modeling. Three widely used approaches in this area, namely Hamiltonian Monte Carlo, the adaptive No-U-Turn Sampler, and Variational Inference, are implemented and compared within a nonlinear regression framework. The analysis is conducted using both simulated and real-world datasets, and the methods are evaluated based on predictive accuracy, quality of posterior approximation, and computational efficiency. The results indicate that sampling-based methods provide more accurate approximations of the posterior distribution, particularly in terms of uncertainty quantification, whereas variational inference, despite a slight reduction in accuracy, offers competitive performance due to its high computational efficiency and rapid convergence. https://andisheyeamari.irstat.ir/article_737336.html Directional data (angular data), including circular and spherical data, have applications in many scientific fields such as ecology, medicine, biology, meteorology, finance, astronomy, geosciences, machine learning, and artificial intelligence. Modeling dependencies among such data has always faced fundamental challenges due to the periodic and nonlinear nature of angles. In recent decades, copulas have gained a prominent place in statistics as powerful tools for separating marginal structures from joint dependence. However, generalizing copulas to directional data—including circular–linear, circular–circular, and spherical–spherical copulas—has followed a complex and ongoing research path. In this paper, using a temporal–thematic approach, the historical development, construction methods, applications, dependence measures, estimation methods, and goodness-of-fit tests for directional copulas are reviewed. https://andisheyeamari.irstat.ir/article_737343.html In this article, we define a new family of models, called the beta Marshall-Olkin extended log-logistic family of distributions (BMO-LL), by adding three shape parameters that generalize some well-known distributions in statistics, such as the log-logistic distribution. We obtain estimates of the model parameters using the methods of moments, maximum likelihood, and Bayesian. Finally, we fit some specific models in the new family to a real datasets to demonstrate their flexibility in fitting to the life data. It is found that the BMO-LL model fits better than competing models, so it can be considered as a suitable distribution for fitting to the life data. https://andisheyeamari.irstat.ir/article_737344.html One of the main challenges in multi-instance regression is the simultaneous selection of influential instances within each bag and predictors associated with the response. In this study, the Bayesian multi-instance regression framework based on Shotgun Stochastic Search (MIR-SSS) was applied to analyze data from burn patients, and its performance was compared with two benchmark methods, namely Aggregated Predictive Weighting (APW) regression and LASSO applied to bag-level representations. Within this framework, variable selection is performed using spike-and-slab priors, whereas instance selection is carried out through a logistic model. The methods were first evaluated under four simulation scenarios with different levels of sparsity and variability and were subsequently applied to real data from 2,024 patients admitted to Amir al-Momenin Burn Hospital in Shiraz, Iran. Simulation results showed that the MIR-SSS model achieved the lowest prediction error and high accuracy in selecting relevant variables under the baseline scenarios. However, under conditions of high variability and the presence of outliers, the APW method demonstrated superior predictive performance in terms of prediction error. The area under the ROC curve also confirmed the model's ability to identify influential instances. In the real-data analysis, MIR-SSS outperformed the two benchmark methods in predictive performance and identified several clinical factors associated with patient outcomes. The results indicate that this framework provides an interpretable tool for the simultaneous selection of instances and variables and enables uncertainty quantification in the analysis of complex medical data. https://andisheyeamari.irstat.ir/article_738705.html Describing the relationship between two or more random variables is a fundamental problem in statistics and probability. This paper, which is organized in two parts, provides a historical and structural review of dependence measures from the 18th century to the present. Part I covers foundational concepts: early definitions of independence and dependence in the works of de Moivre, Bayes, and Laplace; the emergence of correlation in Galton's work and its mathematical formulation by Pearson; the limitations of the product-moment correlation coefficient; and the development of nonparametric rank-based coefficients. The role of copula theory and Sklar's theorem in separating the dependence structure from marginal distributions is also explained. Part II addresses modern measures and generalizations: measures of intensity of dependence, non-monotone dependence and modern measures for high-dimensional data, serial dependence and autocorrelation, tail and quantile dependence, multivariate and vector extensions, measures for discrete and mixed data, applications in probability theory, and implementation in statistical software. This paper serves as a comprehensive resource for researchers in statistics, data mining, and machine learning by providing a review of both classical and modern measures. https://andisheyeamari.irstat.ir/article_741590.html ‎The assumption of independence and normality of ‎innovations‎ are usual assumptions in the study of time series models. But in some researches and studies, these assumptions cause limitations, in other words, we encounter cases where ‎the innovations‎ are not independent or do not follow the ‎Gaussian‎ distribution. In this Paper, we consider ‎autoregressive‎ model with dependent and non-‎Gaussian ‎innovations‎. It also is assumed that the true model is unknown, so we propose some competing models. The unknown parameters of competing models are estimated using the modified maximum likelihood method. Finally, by using information criteria, such as Akaike's information criteria and model selection tests such as Vuong's test, the optimal model has been selected. Using simulation, the performance of the modified maximum likelihood method has been studied and it has also been shown that the model selection criteria and tests select the closest model to the true model as the optimal model.