fuzzy clustering

Nowadays, the organizations’, especially banks’ interaction with customers has changed significantly as customer relationship management (CRM). The current research explains a model for customer relationship with a Fuzzy-Bayesian recommender approach. It is an applied and exploratory research, which has reviewed the information of 98,604 customers. It has used 3 approaches of data mining, fuzzy, non-linear Bayesian averaging. The information of 22 indicators of customers was inputted into non-linear Bayesian models (BMA, TVP-DMA and TVP-DMS). The BMA model, regarding its error rate, had the highest accuracy. As the results show, account balance, total balance of deposits, total balance of current facilities and volume of financial transactions were the research’s non-fragile variables. C-MEANS approach is more accurate than K-MEANS. The C-MEANS approach could identify 16 clusters and analyze the characteristics of each of 16 clusters. Thus, most of the variables have a positive and significant effect on the status of each cluster.

Edges are one of the most important parts of an image. The border between two areas of the image that has a significant difference in brightness, color or texture is called an edge. Edge detection is a basic step in image segmentation and one of the most important steps to detect the characteristics of objects in an image. Recently, many researches have been carried out in the field of edge detection, each of which has tried to better detect the edges of the image. In this research, fuzzy theory and the concept of fuzzy clustering are used to detect the edge of digital images. At first, the pixels of the image are based on the amount of gray level. They are divided into clusters and then by determining the appropriate filters, the edge of the image is detected. The proposed method has recognized the edges of the image well, and has also shown remarkable performance in identifying the edges of the image that cannot be correctly recognized due to the inappropriate light level or noise. The results of this research showed that the proposed method has provided better results than usual methods in most cases and this is due to its good performance in detecting image details such as thickness and strong and weak edges, especially in images with lower resolution or lower contrast. It has performed very well.

Nowadays, quality management is an important field for organizations, including the banking system, which faces fierce competition. Organizational Excellence Models serve as guides to help organizations improve their quality. The main aim of this study was to investigate the dynamic relationships between the approaches and criteria of the EFQM organizational excellence model and optimize the simulation of these relationships. To do this, we first identified the common approaches in the organizational excellence model and clustered them using the fuzzy clustering method. Next, we used the systems dynamics method of causal rings to discover the dynamic relationships between these approaches. Finally, using data from the banking system in Iran, we simulated cases and optimized the growth rate of organizational excellence in the sample over 20 years (2018-2019).

Nowadays, updating information collected from urban areas is of great importance, since it provides the basis for many fields of study such as land cover changes and environmental studies. Remote sensing provides an opportunity to obtain information from urban areas at different levels of accuracy while widely used in various change detection applications. Detecting changes in buildings as one of the most important features in urban areas is of particular importance. Powerful and expensive processing systems are the only way to process large volume of remote sensing and photogrammetry data generated by the ever increasing number of sources to which laymen do not have access. The present study has applied deep learning methods and high computational volume of data processing in free clouds to make this possible for the public.

Two case studies have been selected in the present study. The first includes DSM and Orthophoto images captured by drones from Mashhad in 2011 and 2016. DSM and Orthophoto images in the second case study has been collected by drones from Aqda in Yazd province in 2015 and 2018. In accordance with the type of data used and high computational volume used for processing, the present study has applied fuzzy clustering method to detect buildings with a high computational speed and deep learning method to detect their changes. Object-based method and fuzzy logic theory have been used in the first step to classify features and detect buildings. In the second step, deep learning method and DSM differentiation method were also used to detect changes in buildings and evaluate results obtained from deep learning method. In the first step, buildings were detected using descriptors extracted from terrestrial and non-terrestrial features, and related decisions were made using fuzzy logic. In the second step, DSM differentiation method has applied the masks extracted from buildings in both epochs on the related DSMs to find their difference and detects changes using an elevation threshold. In deep learning method, a convolutional neural network model was trained to detect changes in buildings during both epochs. Using the DSM of buildings in both epochs and a part of their interface, the network input layers were generated for training. Changes detected in the buildings by the differentiation method were also introduced as the output layer. Following the training and introducing the entire interface in both epochs as the input layer, the trained neural network has detect changes in the buildings. The same process was performed once more using the difference between two DSMs. In other words, a single input layer was used in the network and the rest of the process was the same as before. Finally, changes detected by the neural network was compared with changes detected in the DSM differentiation

In the first step, buildings were detected and images were classified in accordance with the fuzzy logic. The overall accuracy of the first epoch classification in Mashhad equaled 94.6% indicating higher acuracy of object-based methods as compared to pixel-based methods. The overall accuracy of first epoch in Aqda equaled 95.5%. Neural network method detected changes in buildings with an overall accuracy of 90%. In accordance with the ground truth used in network training (both using DSMs as the input layer and the difference between the epochs as the input layer), results indicated that deep learning method is highly accurate in one-dimensional convolution mode. Moreover, the second step has applied the difference between DSMs in the two epochs and thus, many areas lacking a change in height were removed in both epochs and the network was trained more appropriately and accurately.

Necessity of extracting features, especially urban features such as buildings and identifying their changes over time have been investigated in the present study. Due to the high computational volume of modern remote sensing and photogrammetry data and highly expensive systems required for their processing, a new method was presented in the present study to solve this problem. Considering the type of data used and the complexity of features, object-based methods were selected instead of pixel-based methods to identify features and buildings. Deep learning method was used to detect changes in buildings. The method was also compared with DSM differentiation method. A one-dimensional convolutional neural network was used in the deep learning method. Two different modes were used in the network to train and predict changes. In the first, DSMs extracted from the buildings in each epoch were used as the input layer, while in the second one, the difference between DSMs were introduced as a single input layer to the network and the network was trained in accordance with the ground truth collected from areas with and without change obtained from the DSM differentiation method. Following the training process, changes were predicted using the trained network. Much better results were obtained from the second mode in which the difference between DSMs were used.

Today, one of the best tools available for value chain engineering is applying the practical lessons of business models. By analyzing a business model, the processes of development, presentation and receipt of value by a company can be improved. Although many companies operating in different industries of the country have increasingly turned to the analysis of their business models, the issue is moving forward at a lower speed in the pharmaceutical industry. Numerous new business models have been developed or renovated in today’s changing business environment. The development and renovation of new business models are not a matter of choice but a necessity in order to survive in an environment whose rules of the game are constantly changing. In such a situation, the greatest challenge posed to organizations is developing a framework through which an accurate understanding of the dimensions and components of their business models can be achieved and a point of reference can be offered to compare their business models with those of the competing companies to enable innovation and outperformance. The main focus of this study is to find a practical framework for pharmaceutical companies so as to identify and compare the key features of their business models with those of active competitors, hence to understand the strengths and positive qualities of other models and overcoming their weaknesses.In this study, the latest scientific findings about the business models of pharmaceutical companies are presented, and the common types of these business models are identified using taxonomy (experimental typology) to achieve a more comprehensive and accurate view of the dynamics of this business model in companies. The statistical population of this study included all the Iranian companies producing human drugs that have had a competitive product in the growth stage. The study was conducted in two phases, content analysis and quantitative phase. Initially, semi-structured interviews during the content analysis phase revealed a comprehensive framework for the dimensions and components of the business model of Iranian pharmaceutical companies, and then, through quantitative surveys and in-person questionnaires, the business model of each pharmaceutical company was identified. The sampling method in the qualitative part of the research was judged after 19 theoretical saturation interviews, and the quantitative research became available. By using the PASS software, an optimal sample of 40 companies was adopted. In this study, the empirical classification (taxonomy) approach, rather than the ideal classification (typology), was used to identify various business models. In clustering methods, since the data are classified irrespective of any presupposition or predetermined framework, the fuzzy c-means (FCM) clustering algorithm serves as the most appropriate tool to attain the typological objective of this study. Finally, by fuzzy clustering with the MATLAB software, four distinct variants of the business model were identified for the pharmaceutical companies, each having significant differences in the key features of the business models with the other types. These four types included Diverse Effectiveness, Concentrated Effectiveness, Diverse Effectiveness, Concentrated Effectiveness. The identified types have specific similarities and differences in each business model dimension and component which can be concisely compared as follows based on their key features including concentration on quality rather than on efficiency, extent of relationships in various areas such as sales, marketing, and raw material supply. The competitive pharmaceutical products in Iran can be categorized in four distinct types in terms of the characteristics of their business model. These models have different characteristics in terms of the amount of value proposition, the processes of shaping the value architecture, the mechanisms for interaction with different groups of customers and value acquisition methods. Finally, these types of business models were named based on their proprietary features as: diversified effective, focused effective, diversified efficient and focused efficient. The first business model type was labeled “diversified effective” as it is greatly focused on quality and extensive relationships with various stakeholders of the value chain. Companies such as CinnaGen, Exir, SobhanDarou, DarouPakhsh, Osvah, and Samisaz had the greatest degree of membership in this business model type. The second type was labeled “diversified efficient” as it is highly concentrated on cost control and extensive relationship maintenance with the value chain stakeholders. Companies such as ParsDarou, TehranShimi, AlborzDarou, TehranDarou, and Actoverco had the greatest degree of membership in this business model type. The third type was labeled “focused efficient” and included such pharmaceutical companies as Hakim, CaspianTamin, Sepidaj, Loghman, and Jalinous which had the greatest degree of membership in this business model type. Finally, the fourth type was labeled “focused effective” and involved companies such as Shafa, Ronak, DinehIran, and Modava with the greatest degree of membership therein.

Recognition of socio-economic characteristics and decision-making and behavior of different families is one of the important tools for evaluating the social policies. So far, family economics have been evaluated based on family welfare criteria in the West and have never been viewed from the perspective of Islamic economic indicators. In this study, based on the gradational family model and the scale designed for it, the researcher has conducted a survey on 387 households in Tehran. In analyzing the results, use was made of not only the descriptive statistics, but also of such inferential statistical methods as the linear regression analysis. Use was also made of the fuzzy C-means clustering method to divide families into different clusters based on the advancement and decline components. Results have shown that the absolute income and assets are not good predictors of the family status and its variations; in contrast, the feeling of economic adequacy and satisfaction can, as a mental and cognitive component, explain a higher percentage of families’ changes and diversities. Although families are relatively well-off in terms of the growth criteria (cooperation, kindness, and transcendence), a high percentage of them experience average naturalness and moderate to high exploitation. Hence, it can be concluded that, while paying attention to transcendent values, families in Tehran have serious weaknesses in some components.

Nowadays, entering the field of quality management is one of particular importance to organizations. This is due to the intense competition, globalization and diversified product services that have made the need for quality development at all levels in organizations inevitable. The country's banking system is no exception because of the fierce competition. Organizational Excellence Models are the tools that guide organizations along this path. Organizations including the country's banking network, are adopting a variety of approaches within their organization to realize the values of excellence models. However, the limitation of available resources such as cost, manpower, organizational culture, etc. does not allow all of these approaches to be deployed. Therefore, the key question is which approaches will have the most impact on quality improvement.

The main purpose of this study is to explore the dynamic relationships between approaches and criteria and sub-criteria of EFQM organizational excellence model. For this purpose, first, the usual approaches in organizational excellence model are extracted and due to their high volume, they are clustered by fuzzy clustering method and finally, causal loops between variables are extracted by system dynamics method.

The result of leading research is the causal linkage model between the categories of common approaches in the organizational excellence model.

In this study, common approaches in I.R. Iran’s banking network are identified and during of using system dynamics, it is possible to simulate the behavior of the system using causal loops.

Talent management is one of the most important challenges faced by management of companies. Uncertainty in predicting competencies that are critical for companies and the inability to accurately estimate the voluntary of talents turnover, has required adapting concepts from the supply chain management, such as supply and strategic sourcing in talent management. The purpose of this research was to cluster the job occupations of the National Iranian South Oil Company (NISOC) in order to provide suitable strategies for sourcing talents in each cluster. This study is considered a quantitative applied research. The research strategy was c-means fuzzy clustering method. The statistical population was all human resources staff in NISOC and the statistical sample consisted of 80 experts from human resources section which have been selected by purposive method. Based on the findings, the job occupations were categorized into four clusters of routine, leverage, bottleneck and critical occupations, and suitable strategies for supplying and sourcing talents for each cluster were proposed.

Various chemical compounds are used in industry. Many industries maintain the results of chemical compounds. In this case, maintaining and using existing chemical data poses a challenge. If the amount of this chemical data increases, a model for data clustering is needed to be able to separate the data of different compounds. Clustering Finds data with similar properties in separate clusters without having prior knowledge of the available data. In the chemical industry, it is not possible to label all chemical composition data because they may occur or change at any time. In this case, clustering should be used, which divides the chemical data into a number of subsets. From a data mining point of view, chemical data detection is one of the issues of data marginalization. By introducing appropriate algorithms in this field and then trying to increase the efficiency and accuracy of chemical information, we can take steps to create highly reliable mechanized systems with the ability to detect complex patterns. Here, a series of data on the chemical composition of different industries are collected and clustering is performed with the help of a suitable hybrid model. The proposed method is a hybrid model of the nearest neighborhood using fuzzy clustering. In this model, the existing chemical data is subjected to a preprocessing operation to remove inappropriate and empty data from the system. The clustering operation is then performed with the nearest neighbor model.

Various chemical compounds are used in industry. Many industries maintain the results of chemical compounds. In this case, maintaining and using existing chemical data poses a challenge. If the amount of this chemical data increases, a model for data clustering is needed to be able to separate the data of different compounds. Clustering Finds data with similar properties in separate clusters without having prior knowledge of the available data. In the chemical industry, it is not possible to label all chemical composition data because they may occur or change at any time. In this case, clustering should be used, which divides the chemical data into a number of subsets. From a data mining point of view, chemical data detection is one of the issues of data marginalization. By introducing appropriate algorithms in this field and then trying to increase the efficiency and accuracy of chemical information, we can take steps to create highly reliable mechanized systems with the ability to detect complex patterns. Here, a series of data on the chemical composition of different industries are collected and clustering is performed with the help of a suitable hybrid model. The proposed method is a hybrid model of the nearest neighborhood using fuzzy clustering. In this model, the existing chemical data is subjected to a preprocessing operation to remove inappropriate and empty data from the system. The clustering operation is then performed with the nearest neighbor model.

Nowadays the growth in the use of social networks among different classes of world community is increasingly undeniable. Social networks database include Rich and valuable resources whose release and analysis with the purpose of marketing, publicity, National Security, Health and etc. can benefit researchers of public and private institutions. But respect the privacy of the entities whose information is available to data miner analysis is essential as a legal protocol. In this Paper, through qualitative methodology Meta synthesis, all related dimensions, indicators and codes were identified and then the importance and priority of each of the factors was determined. Subsequently, the improved model of anonymity was presented by an optimizing firefly algorithm and fuzzy clustering. The result of simulation and assessment of the proposed model on the data of four social networks such as Facebook, YouTube, Twitter and Google+ depicts that privacy preserving of data with the lowest distortion ratio and the more usefulness of data.

The aim of this study is clustering of Iran precipitation in order to recognize its spatial differences. Daily data of precipitation, relative humidity, and due point temperature were acquired from 63 stations since the year 1980 till 2013 by national meteorology organization. These primary data by using of the method Singular Value Decomposition (SVD) mapping were compressed to single values and then decreased and prepared as non-linear properties. In addition to these properties, some other properties such as the days with heavy precipitation (greater than or equal to 10 mm), days with much heavy precipitation (greater than or equal to 20 mm), days with precipitation greater than or equal to 25 mm, greatest number of consecutive dry days (CDD), greatest number of consecutive wet days (CWD), and the number of days with precipitation at every season were acquired as linear properties of daily precipitation. In the final step, these linear and non-linear properties were joined together then entered to the fuzzy clustering system. The result of this study showed six precipitation clusters in Iran. The coastal areas of the Caspian Sea, Persian Gulf and Oman Sea, the central very dry, semi-arid, mountainous and semi-mountainous areas were grouped in definite clusters according to Iran topography, Latitude and distance from the sea and water resources. Comparing the results of clustering by this method with other methods that have been done until now, we found that this new approach of clustering can distinguished the stations and clusters regarding its causative agents, significantly high performance.

Remote sensing method known an appropriate tool for estimating depth in the coastal environment of the limited reaches. The purpose of this study is to measure the depth of the southeastern coast of the Caspian Sea through the training of the neural network. In order to estimate depth, Landsat-8 images and hydrographic data collected using the echosounder, have been used. Atmospheric correction of Dark Object Subtract (DOS), radiometric correction (turning digital number to reflection), the sun glint correction, and eventually masking the water body from the land area, applied on the coastal blue, blue, green and red bands. These steps known as pre-processing.
In this study, depth estimation through the neural network is investigated in two states. In the first case, each of the four bands as input and real depth corresponding to each of these pixels as target was introduced to the neural network. In the second case, the depth data were clustered to seven clusters by the fuzzy C-mean (FCM) method. After clustering, the data of each cluster was separately presented to the network. In both cases, the share of train data, validation data and test data from input data is 60%, 10% and 30%, respectively. The results of the neural network indicate that the accuracy of the estimated depth in various clusters is different, and the highest accuracy (R2 = 0.90, RMSE= 0.11) and the lowest accuracy (R2 = 0.67, RMSE= 0.11) belong to cluster (1) and cluster (3) respectively. As well as, the estimated depth in no clustered data, evaluated with high accuracy )R2 = 0.98, RMSE = 0.16).Then, neural network method is able to estimate depth from shallow coastal waters with high accuracy.

Objective There is a general tendency toward direct marketing these days. Therefore, instead of designing advertisement and marketing strategies for all the customers in the market, it is recommended to classify the customers based on clustering techniques and then design specific strategies accordingly. This will reduce marketing and advertisement expenses, increase sale department efficiently, build closer and quicker relationships with different customers and etc. There are a variety of clustering methods. Provided that clustering means classifying customers in different groups with maximum similarities within the groups and maximum difference among the groups, it may not be appropriate to apply such a rule in clustering customers (people) due to their nature. Hence, fuzzy clustering technique seems more appropriate for customers because there are no absolute borders considered among different groups just as the market suggests. This study, then, aims to emphasize on this concept in order to apply fuzzy clustering on market.

Methodology This practical research is descriptive-exploratory in nature of data collection. The statistical population includes all the customers of a mobile company, but due to availability issues only a part of their customers would be involved in the present study. A questionnaire including 6 questions was distributed among those customers and only 760 were correctly responded. Finally, EXCEL and S-PLUS were used to analyze the data. Findings The data in this study include three different parts of information. The first part includes some indexes selected for analysis of the clustering. Second part concerns with the customers service usage such as distant phone calls, free calls and wireless services. The third part refers to other mobile services provided for each customer. This part is
presented in a binary fashion deciding whether a customer has received a specific service or not. Such services include activating more than one mobile line at the moment, using voicemail, paging, internet and other services. This algorithm was used to conduct fuzzy clustering in the present study. Following applying fuzzy clustering, only 2 clusters were judged appropriate for such data. The first cluster includes customers with lower income, job stability and lower loyalty to the mobile company, while the second cluster includes customers with higher income, higher job stability and more loyalty to the mobile company. The customers in the first cluster used services like free calls, wireless networks and pay phones. On the other hand, the customer in the second cluster mainly used services like distant calls and rarely used wireless services. In general, we can claim that paging services were the highest requested and then there are voicemail services, internet, and e-pay services respectively. The two clusters reported to have a similar tendency in using services such as voicemail, multi-lines, conferencing; yet, they were different in services like paging, internet, call forwarding (diverting), call waiting and epay services. At the end, mobile companies can set marketing strategies based on such findings.

Conclusion It is suggested that mobile companies focus on general advertisements and distant call services, but only a little focus on wireless services. They can also put more thought on services like paging, voicemail, internet and e-pay services respectively. It is also recommended that, for female customers (mostly within the first cluster), the companies should focus on pay phone services, distant calls, and free calls as well as voicemail and internet. On the other hand, for male customer with higher job stability, it is suggested to focus the most on distant call services and provision of special discounts with this regard, but the least on wireless and pay phone services. Besides, voicemail services, paging, call waiting, call forwarding and e-pay services should be the mobile company’s priority for male customers.

Management of watersheds requires understanding of watershed conditions both in gauged and ungaugedbasins. The classification of watersheds by similarcharacteristics for the implementation of coordinated watershed operations and flood control as well as giving priority to sub-basinsis of great importance. The need for a classification framework in hydrology is not an entirely new subject. In fact, this subject has long been discussed and several studies have also attempted to advance this idea. So far, no acceptedcomprehensive protocol has been presented for the classification of watersheds,and questions can be raised regarding why this has not happened. More efforts must be made in order to develop such a classification.Previous studies have used hard clustering methods more, for the classification of watersheds but, the present study used fuzzy approach as asoft method. In general, the purpose of this research is to focus on the characteristics of the watersheds including morphological characteristics, soil and land use for the identification of similar watersheds. These parameters can facilitate the watershed classification scheme and our understanding ofthe watershed conditions. The dataset for this study includes is base maps (sub-watersheds boundary, streams and rivers, digital elevation model (DEM), soil and landuse which have been collected from Iran Water Resources Management Company. To classify the Karkheh watershed, 35 spatio-physical indices including topographic, morphological, landuse characteristics and soil parameters were considered. These indices have been calculated for each watershed. The dimension reduction of the variables was an important part, because 35 indices were quite large for the classification of 38 watersheds. Therefore, factor analysis for each group of indices wasusedseparately to reduce the number of variables.
After reducing the variables and selecting the final indices, the fuzzy clustering approach was conducted to classify the watersheds into homogenous groups. The number of optimal clusterswas determined through trial and error and the functions of partition coefficient and partition entropy evaluation. Kaiser-Meyer-Olkin (KMO) test statistics for each group of the morphological, landuse and soil indices were 0.71, 0.69 and 0.76 respectively, indicating that the data was suitable for factor analysis. Factor analysis was conducted using Principle Component Analysis (PCA) method and the results revealed that among 35 spatio-physical indices, 9 indices (4 morphological indices, 3 land use indices and 2 soil parameters) had a higher load factor than other indices. Therefore, indices of the watershed surface, basin elongation, average length of drainage network and total topographic indexamong the morphological indices;percentage indices of rangelands, agricultural lands and wastelandsamong the land use indices; and indices of water holding capacity in the soil layer and saturated hydraulic conductivity among the soil parameters were selected as the ultimate criteria for grouping the watersheds.
Theselectedfactors were normalized between zero and one before the classification. Then, sub-watersheds were classified using fuzzy C-mean (FCM) approach. The trial and error method was used to find thenumber of optimum clusters. The maximum amount of evaluation function of partition coefficient equal to 0.76 and the minimum amount of partition entropy function equal to 0.49 occurred in three clusterstherefore,the number of optimum clusters equal to 3 clusters was determined through trial and error.The results of classification indicated that the triple groups included the sub-watersheds of the northeastern regions and parts of central regions of the Karkheh watershed (group 1), the northwestern- southeasternalong with the southern regions of Karkheh watershed (group 2) and the central regions and parts of southwestern regions of Karkheh watershed (group 3). Watershed classification with similar characteristics can be used as a method for watershed management, flood control and giving the priority to critical sub-basins. However, watershed classification is only completedwhenit is understood why some catchments belong to certain groups of hydrological behavior, so as to be possible to classify gauged and ungaugedwatersheds through it.
Finally, it is important to remember that classification of watersheds is not the “be-all and end-all” of research on watersheds, but rather only a means towards achieving broader aims of planning and management of our ecosystems, environment, water resources, and other relevant earth systems and resources. However, watershed classification certainly allows us to study catchments more effectively and efficiently and develop more appropriate strategies in terms of simplification in models/model development, generalization in our modeling approach, and improvement in communication both within the hydrologic community and across disciplines, as much as possible.

Iran's pharmaceutical industry has long been confronted with various marketing and advertising constraints Most of these constraints has arised from governmental terms and conditions, and has led to the overwhelming majority of these companies pursuing a passive and minimalistic approach to exploiting promotional strategies. However,today, with the arrival of newly stablished pharmaceutical companies and with private sector support, we are observing a remarkable change in the past approaches of these companies toward their promotion strategies.
In this research, presenting the latest findings of promotion strategies of human pharmaceutical companies, we aimed to identify the common types of these strategies using the taxonomic method. For this purpose, we first created a comprehensive framework for the dimensions and components of the promotion strategy of pharmaceutical companies by conducting semi-structured interviews in a qualitative research and using a content analysis method ,then, through a quantitative survey and completion of questionnaires, the promotion strategy of each of the forty pharmaceutical companies in the statistical sample of this study was identified. Finally, by performing fuzzy clustering, four clusters or distinct types of promotion strategies of pharmaceutical companies were identified, each with significant differences in some of the key characteristics of other types. A key result of study shows that pharmaceutical companies have adopted different approaches to using promotion strategies

Evaporation is the important factor that affects temperature, drought severity and water storage in the hydrological cycle and plays an important role on managing the water resources projects such as agricultural irrigation.
Classification of datasets is useful for concisely system modeling purposes. By classification, a large number of datasets is reduced to a small number of groups. In the field of hydrological systems, classification of meteorological stations into homogeneous groups will be useful to consider a different scale of measure, which is suitable to each group. Such classification can lead to choice methods appropriate for each group for management of water resources in various regions. Classification will also be useful for prediction of events such as droughts. Moreover, in the case of estimating missing data, the corresponding data of the representative station determined using a classification technique can be successfully substituted (Raju and Kumar 2007).
stations. Dikbas et al. (2011) applied the FCM method to classify the precipitation series and identify the hydrologically homogeneous groups in Turkish. Regional homogeneity test results showed that regions determined by the FCM approach are sufficiently homogeneous for regional frequency analysis.
In the present study, the practical applicability of two classification methods, namely fuzzy c-means (FCM) cluster analysis and Kohonen artificial neural networks (KANN), is examined for grouping 97 evaporation stations in Iran into homogeneous groups. The rest of the paper is organized as follows. First, a description of the case study is presented. After introducing the applied methods, results obtained are presented and discussed and conclusion drawn.

One of the major challenges of customer-centric organizations is the recognition of customers, the distinction between different groups of customers and their ranking. Clustering is one of the data mining techniques used to group customers into their various characteristics. The main purpose of the research is to customer clustering based on the Recency, Frequency and Monetary indicators using the fuzzy c-means algorithm. The study was conducted on 76379 registered transactions from customers of Zahedan City Refah Chain Store. The results of this research provide a framework for developing customer relationship management programs for each customer group.
Introduction Today, the importance of customer relationship management is not hidden from anyone and all service and product companies are trying to understand more of their customers. Understanding the various groups of customers and building effective relationships with them in a way that guarantees the economic benefits of companies in the future is an important issue in today's businesses. Maintaining valuable customers and attracting profitable customers is both important and it is possible to accurately identify their features. Clustering is one way that helps companies recognize their profitable customers. In the clustering of the elements within each cluster, the most similarities are found, and there is a significant difference between clusters. By introducing the fuzzy theory by Lotfi zadeh, the application of this idea in various sciences quickly expanded and the fuzzy clustering method was widely used by researchers in various fields (De Oliveira & Pedrycz, 2007). In fact, the main difference between the classic clustering and fuzzy clustering is that an instance can belong to more than one cluster (Khoshnazar, 2013). Companies with customer clustering and behavior analysis of each group will provide a platform for optimal allocation of resources and developing customer relationship management strategies. The customer lifetime value (CLV), reflects the value that can help companies in this field. Customer lifetime value is the value of the customer creates throughout his lifetime and is determined by using different models (Boroufar, Rezaeian & Shokohyar, 2017). The RFM model is one of the most popular and effective methods for analyzing customer life value. This model uses three variables Recency, Frequency, Monetary to express the difference between customers and the customer lifetime value is calculated from the sum of the values of the model's indexes. It is also assumed that customers who are worth a lot on any of the model's indicators are the most profitable customers. Of course, they will behave like they were in the future.
Case study Rafah Chain Store Company is one of the most comprehensive distribution networks in Iran with the aim of supplying and distributing basic goods.
Materials and methods In this research, transactions recorded in the database of the Zahedan Refah chain store have been used in a seven-month period. After receiving the data and performing the preparation process, 76,379 transactions were used as the final input. The preparation process consists of two steps. In the first step, the data was cleared, so some data with invalid values were identified and deleted. In the second stage, RFM model indices were calculated using SPSS Modeler 18 software. There is a difference in the RFM model index unit so these values should be normalized to the same unit. For this purpose, these values were normalized using the Min-Max method. To determine the number of clusters, the Xie and Beni index were used. After calculating the value of this index, 7 clusters were determined as the optimal number of clusters. Fuzzy C-means algorithm is used to cluster customers based on RFM model indicators. All stages of fuzzy clustering and determination of the number of clusters were done using MATLAB software. After fuzzy clustering is done, we will determine the weights of RFM model indices. For this purpose, Fuzzy AHP method was used. Finally, Customer lifetime value for each cluster was calculated from customers and clusters were ranked.
Discussion and Conclusion By calculating the lifetime value for each cluster, companies can use their limited resources for a group of customers who have the highest value. According to the results, the fifth cluster with 0.16624 is the most valuable group of store customers. The services provided to this group should not be limited to regular programs, but should be tailor made for them. In fact, the store should allocate more funds to these customers. On the other hand, the third cluster with 0.01482 is the least valuable group of store customers. In developing customer relationship management strategies for this group, there should be a proper balance between the costs associated with the revenue that these customers receive from the store. The results of this research can be used to develop customer relationship management strategies for each customer cluster.

Clustering techniques need to define the number of clusters before they can be applied to the partitioning problem. Determining suitable number of clusters in partitioning problem is the purpose of clustering validity indices, which are nowadays significantly considerable for data miners and this resulted in various numbers of related indices. Separation and compactness information of fuzzy clusters are both considered in developing the advance indices of clusters validity, while this makes the above mentioned indices inefficient because of mathematical sophistication and the need for more computational effort. Therefore, this paper proposes FCI as a new index, which employs fuzzy cardinality concept in defining the number of clusters in fuzzy clustering. FCI also considers both compactness and separation of fuzzy clusters while significantly decreases computational efforts. In this paper, after reviewing the cluster validity indices and fuzzy clustering algorithms, FCI index will be explained and ultimately to evaluate its effectiveness will be implemented.

One of the most important issues in Customer Relationship Management is customer segmentation and product offer based on their needs. In practice, Customer’s behavior will change over the time by changes in technology, increase in the number of new customers and new competitors, and product variety. Traditional segmentation models that are static over time cannot predict these changes in customer’s behavior and ignore them. This challenge is especially critical in Telecommunication with high churn rates. In this research, we have used temporal fuzzy clustering to detect significant changes in customer's behavior for a telecom company during a 10-month period. The aim of this study is to find factors that affect structural and gradual changes in clustering model. In addition, we have suggested a method based on Frechet distance to extract similar patterns in customer’s usage behavior. Provided that combining the temporal clustering with trajectory analysis is an effective way to recognize customers’ behavior among the clusters, the results showed that there are seven distinct customer behavior patterns two of which lead to the customer drop or churn. These patterns can be used to reduce the risk and costs of customers churn and to design optimum services.