bayesian network

Disruptions can impact not only suppliers and manufacturers but also influence each other at the beginning of the chain or customer demand at the end of the chain. In this paper, the extent of this impact is modeled and solved using a Bayesian network. Inflation rate is used to predict and reduce demand uncertainties in a linear programming model with two objective functions of increasing geographic dispersion and reducing total cost (transportation, purchasing, ordering, etc.). In this model, suppliers and manufacturers collaborate to increase supply chain resilience. For the first time, the concept of supplier resilience level is proposed. The proposed model for order allocation, in addition to price and other ordering costs, also considers the cost of improving the resilience level of suppliers. Also, customer satisfaction level is implicitly increased by increasing the cost of unmet demand. To this end, a case study was conducted in one of Iran’s automotive companies. To validate the proposed model, a numerical example was solved and sensitivity analysis was performed. To reduce the number of scenarios, fuzzy c-means clustering and balanced interaction analysis were used. The proposed model can prepare manufacturers for better decision-making and planning in the face of future risks and uncertainties.

The COVID-19 pandemic has posed a significant challenge to the COVID-19 vaccine supply chain that need to be resolved for a successful exit from the pandemic. It is essential to identify and evaluate the risks associated with the vaccine supply chain. We propose a new measure for quantifying the resilience of the vaccine supply chain in Iran. To evaluate the resilience of the vaccine supply chain, we use a Bayesian network to consider the vulnerability and recoverability indices as well as the associated disruption propagation. This approach enables managers to measure the resilience of the supply chain and identify the reasons for performance decline due to the ripple effect. Our results show that outbreak risks, lack of access to vaccine suppliers, low efficacy of the vaccine against new variants, inaccurate prediction of vaccine demand, and failure to choose the right suppliers are the risks likely to have a high impact on the disruption of the vaccine supply chain. Our approach provides a useful tool for evaluating the resilience of the vaccine supply chain and identifying the critical risks. It can be used by decision-makers to mitigate the impact of disruptions and improve overall resilience of the vaccine supply chain.

Today, companies and organizations are aware of the fact that customer retention leads to greater profitability. Increasing competition causes the rate of customer churn to grow. Therefore, studying the features influencing the tendency of customer churn is important. In the present study, a hybrid model based on the data mining approach is presented to analyze the features of churn customers. In the first step, the feature selection node has been used to identify the features with higher importance and remove redundant items. Then, the C5.0 Decision Tree and Bayesian network were used to classify the customers into two groups, turning customers, and non-turning customers. These are data mining techniques and terms that can help in forecasting. Finally, the proposed model has been implemented in the chain store industry as a case study. Key findings indicate that both the decision tree model and the Bayesian network can predict churn customers with different accuracies, the area under the receiver operating curve in the decision tree model is greater than the Bayesian network model and it has better performance. The results indicate the optimal efficiency of the proposed method. In addition, the results show that three features of gender, marital status, and age from the set of demographic characteristics and five factors of average monthly income level, number of purchases per month, the share of online shopping, how to get to know the store, type of market-related to customer transaction records are among the most effective factors.

In light of the continuous and rapid changes in global competition, companies face the imperative of consistently introducing new products or expanding their existing product lines to maintain their competitive edge. Recognizing that numerous factors within the supply chain influence the production, design, distribution, and introduction of new products, understanding supply chain risks is crucial, spanning from the procurement of raw materials to the delivery of products to the market. Consequently, risk management stands as one of the most critical challenges within the supply chain, significantly impacting New Product Development (NPD) performance. This research seeks to answer the primary question: "How and to what extent do various supply chain risks affect newly developed products?" While prior research has employed various methods to evaluate and manage supply chain risks, few models have explored the interplay of these risks on each other and their influence on performance dimensions. In this study, based on a review of theoretical foundations and prior research within the clothing manufacturing sector, we identified dimensions of newly developed products and supply chain risks. We employed the Delphi technique through interviews to identify the most significant risks. Subsequently, we employed the Cross-Impact Analysis method to elucidate relationships between these factors. Finally, we utilized Bayesian networks to analyze the impact of identified risks on the performance of the selected new product, conducting sensitivity and scenario analyses. The findings indicate that environmental and supply risks are more likely to manifest than other risks, with three operational, distribution, and demand risks, influenced by environmental and supply risks, exerting the most direct impact on new product performance, particularly in the dimension of quality.IntroductionModern organizations recognize that traditional competitive strategies, such as improving quality and reducing costs, no longer suffice to remain competitive. Research has demonstrated that numerous new product development NPD projects face failure for various reasons. Effective risk identification and management, particularly concerning supply chain risks in NPD projects marked by a high degree of uncertainty, emerge as pivotal factors for NPD success. In this context, the clothing sector, characterized by a complex supply chain structure, has been extensively studied. However, prior research has predominantly examined existing risks individually, overlooking the interactions between risk components and their simultaneous effects on one or more project objectives. In this research, we not only assess the simultaneous impact of risks on product performance using the Bayesian network method, an effective approach in supply chain risk analysis, but also investigate the severity of risk impacts under different scenarios. This research addresses three primary

Identification of supply chain risks in the clothing industry based on background research and case studies.Determination of interdependencies among variables using conditional modeling.Evaluation of the influence of supply chain risks on new product performance using the Bayesian network method under varying scenarios.Literature reviewNumerous researchers have investigated supply chain risks and their repercussions on product and organizational performance. Asgharenjad Nouri et al. (2021), in their article titled "The Effect of Risk Management on New Product Development in the Banking Industry," explored the impact of various risk indicators on new product development. Their results underscore the significant positive influence of managing all risk indicators, including technology, market, environment, finance, organizational resources, and commercialization, on new product development. Qazi et al. (2017), in their article titled “Supply Chain Risk Network Management,” prioritized risks and corresponding strategies through a case study involving semi-structured interviews. They initially identified organizational performance criteria and then linked them to relevant risks, using a matrix of expected profit to investigate the impact of risks on specified performance criteria. Subsequently, they employed the "weighted net evaluation" method to assess practical strategies.MethodologyIn conducting this research, we initially extracted supply chain risks and product performance dimensions from the existing literature. Subsequently, we employed the Delphi technique to select the most significant supply chain risks, providing indicators to participating experts through questionnaires with a 5-point Likert scale. We then used the Content Validity Ratio (CVR) index to confirm or reject the components derived from the questionnaires. In the next step, we used the Cross-Impact Analysis method, employing pairwise comparisons via questionnaires, to reveal relationships between the key risk criteria. Finally, we investigated the impact of identified risks on the performance of the selected new product within the supply chain of Happy Land factory using the Bayesian network method under various scenarios.Discussion and ResultsThe results from the Bayesian network analysis in this research demonstrate that environmental risk, as an external risk within Happy Land’s supply chain, exerts the most significant influence at the highest level of the Bayesian map. Subsequently, other risks, including economic risks, supplier risks, distribution risks, operational risks, and demand risks, are categorized in subsequent levels. Additionally, sensitivity analysis scenarios, depicted in the Tornado chart, reveal that supply chain risks have a substantial impact on performance criteria. According to this scenario analysis, the primary risk affecting quality and cost target nodes is operational risk, while the major risk affecting the product delivery time node is distribution risk, and the primary risk influencing profitability is demand risk. Results from both pessimistic and optimistic scenario analyses under the second scenario of the research indicate that in the pessimistic state, the presence of a high percentage of these risks significantly negatively impacts quality performance. Conversely, in optimistic scenarios, where these risk factors are not present, improvements in quality's functional dimension exhibit the most substantial impact.ConclusionWhen introducing a new product to the market, evaluating and managing supply chain uncertainties is essential due to the mutual influence of new product development and the supply chain. Supply chain risk management, which commences with the accurate identification and assessment of risks and proceeds with appropriate responses, is crucial for providing efficient and effective new products to the market. In addition to employing the Bayesian network method, a highly effective tool in supply chain risk analysis, we have endeavored to evaluate the simultaneous impact of risks on product performance and assess the severity of risk impacts under various scenarios, including optimistic, pessimistic, and sensitivity analyses. Scenario building proves to be an effective method for validating a developed model to measure the impact of risks under different conditions on target criteria.

The disruption caused by natural, economic and political crises, piracy, and disturbance in information systems and human agents (low probability factors with high impact intensity) travel like a wave from the upstream of the supply chain, i.e. from the suppliers to its downstream. The ripple effect is the phenomenon of disruption in the supply chain with serious impacts on the performance of the supply chain. It can influence sales, on-time delivery, and profits. It has more serious consequences in reducing short-term performance, which leads to market share reduction. This disruption affects the entire supply chain. It destroys the capacity or inventory of the chain, disrupts its facilities, and causes material loss and production reduction in the next stages of the supply chain. Since disruptions are inevitable in supply chains, it is necessary to address this issue and evaluate the impact of disruption risk and its distribution on the supply chain. The purpose of the present study is to provide a simulation model for the behavior of disruption propagation in a three-stage supply chain. It tries to take into account both the vulnerability and recovery capabilities of the disrupted supplier and to determine a criterion for quantitatively predicting the ripple effect of the supplier disruption on the distributor. As the disruption in suppliers prolongs deliveries to customers, and any increase in lead time will decrease sales, the current study seeks to present the quantitative estimation of the ripple effect of supplier disruption on the distributor in terms of lead time and lost sales.
First, using a discrete-time Markov chain, a recovery and vulnerability model was proposed for long-term disrupted suppliers with three states of operational, semi-operational, and fully disrupted. The model was integrated with a Bayesian network to show the way the supplier disruption spreads to the manufacturer and distributor by dynamic Bayesian modeling. Then, in order to quantitatively show the ripple effect of the supplier disruption on the distributor, a criterion was established in terms of the lead time for the delivery of goods and lost sales based on the decision tree and the data of the dynamic Bayesian simulation model.
The model was presented through figures, definitions, and mathematical relations. Its capability was shown by an example based on the status of suppliers, manufacturers, and distributors in the production of a voltage-increasing electronic component. The obtained results showed that when a disturbance occurs in upstream of a supply chain, it spreads as a ripple to its downstream and affects its performance.
The proposed model by this study can quantitatively show the spread and impact of disruption along the supply chain. It can reveal the hidden risk paths and the role of each entity in the chain at the time of disruption. By quantitatively estimating the vulnerability caused by any disruption in the supply chain, the proposed model can help managers identify deviations or the risk of deviations in the chain in time. It can also help them with analyzing and performing control measures to restore the operation and process of the chain and prioritize possible policies and recovery. It is also useful for the selection of suppliers and inventory plans and making the right decisions to reduce the vulnerability of the supply chain sectors quickly with minimal cost.

In today's societies, given the increasing tendency to buy from malls and megastores that can meet all the buyer's needs in a limited time, we witness a large-scale growth of modern malls. Due to the breadth of services and diversity in such places, it is necessary to have a central intelligent control unit that takes control of all existing systems. The Internet of Things is one of the new technologies in the last decade, playing an essential role in intelligent business computing. The volume of customers coming to the mall, the customer service, and the customer buying behavior are factors whose analyses can dramatically increase the revenue from the stores and lead to better customer satisfaction. Many models are used to serve this purpose, one of which is the Bayesian Network model. In this research, using this model based on the interests and preferences of customers and purchasing patterns, we identify customers' needs and provide them with the desired product. After implementing this model, increasing the potential sales in the store and increasing the performance and speed of the store function are expected.

Extended AbstractIntroductionTourism has positive economic, social, and cultural effects on human societies and is one of the most valuable sources of income in many countries. Travel planning is one of the most important issues that can be considered in order to have a good and desirable trip. Travel recommendation systems are examples of techniques used in the field of tourism that aim to match the characteristics of tourism resources or tourist attractions with the needs and priorities of tourists and provide the most appropriate tourism place. Voluntary geographic information (VGI) can play an important role in this regard. People in the community can monitor and share the geographic information of their environment as active, analytical, intelligent, responsible, environment-aware, circulate, distributed and interactive sensors.This study recommends a system that intelligently receives demographic information, users' interests and preferences, and according to the spatial information shared as a VGI by other tourists, recommends the most suitable tourist attractions to the visitors using the recommendation system based on Bayesian networks. Also, the tourism-oriented GIS system, using the TSP algorithm, provides the most optimal route to reach the tourist attractions on the map. MethodologyThe study area is the historical area of ​​Yazd with an approximate area of ​​3.5 square kilometers, which is located in the city center. Yazd city, with an area of ​​107.4 square kilometers, is the largest historical unit and administrative center of Yazd province, which in recent years has faced a very large population growth compared to other urban areas of Yazd province.In this research, the database of Yazd Municipality has been used to collect statistical information on tourism visits in 2015, the vector layer of the regional road network and the vector layers related to the location of tourist places, services and all related facilities. Also, in order to implement the tourism recommender system with a suitable user interface, software programs such as Arc GIS 10.5, PostgreSQL 9.6.1, Visual Studio 2013, NETICA 6.3, Notepad ++, and Geo-Server 2.13.2 have been exerted.Bayesian network is one of the methods of presenting knowledge by combining Bayesian theory with a graphical model and has many applications in problem-solving and cause and effect analysis. This method uses a posterior probability distribution to analyze various parameters and includes a set of nodes and directional edges that nodes represent variables and edges represent cause and effect relationships between variables.The Bayesian network designed in this study has been qualified and quantified based on demographic information (age, gender, income, occupation, level of education and type of travel of users), interests and preferences of tourists, and previous knowledge (information collected from related sources) in the NETICA software.VGI is an essential source of geographic information when data collected from other sources is impossible. The information system designed in this research has a system to enter the preferences and interests of tourists after visiting tourist attractions. In this system, the experience of others is considered more than the experience of new people.In addition to providing a list of tourist destinations tailored to users' preferences, a tourism recommender system also identifies the best route to access each tourist attraction. The routes suggested in this study start by moving from a tourist residence or hotel and return to the starting point after visiting the recommended places. For this reason, the TSP problem has been used for optimal routing in this system. Results and discussionThis article aims to develop a tourism recommender system to offer suitable tourist attractions to tourists. The system can predict tourist attractions tailored to new users' interests and access paths based on combining the demographic information of new users with data from previous ones who voluntarily share their views of tourist attractions by composing recommendation algorithms and VGI data in the form of a Web-GIS system.After introducing the tourist attractions, the user's movement route between the tourist attractions is done by specifying the starting point (Mehr Hotel). Also, in order to move the user, the suggested locations are extracted and the optimal route introduced by the system is displayed on the OSM map with the help of the network layer of the area roads. At any time, the user can view the nearest facilities on the map, such as coffee shops, restaurants, gas stations, hospitals, etc. ConclusionDetermining the appropriate tourist attractions according to the interests and preferences of tourists is one of the important measures in tourism planning. Tourists will be delighted with a trip when their needs, interests, preferences, demographics, and social conditions are considered in the travel planning. In this research, a GIS web-based tourism recommender system has been created using voluntary geographic information and tourist demographic information. The system uses Bayesian network modeling to analyze the preferences of tourists and can predict the behavior of visiting tourist attractions through conditional probabilities and display the impact of each parameter and factor in selecting the type of tourist attractions. One of the most important features of this system is the possibility of user interaction so that tourists can voluntarily share their opinions and suggestions about the visited attractions. So, the system will provide recommendations related to tourist attractions that are in accordance with the interests and preferences of new tourists.

In order to gain more illicit profit, some traders in the stock market try to make a targeted impact on prices by placing fake orders and false advertising. Due to the high customer population, it is not possible to discover these frauds using traditional methods. The present study seeks to provide a system for preventing the frauds in future market-trading coins based on Bayesian classifier model for Iran Mercantile Exchange. The proposed model has polynomial time complexity and high accuracy because of considering important dependencies among different features of data. The primary labeling of data has been done by Kmeans clustering. The test of model shows 94.55 percent similarity between model's output and labeled data. Using this system can helps to identify the fraudulent from non-fraudulent traders.