Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5, 1576-1592 2025 Publisher: Learning Gate DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate © 2025 by the author; licensee Learning Gate History: Received: 17 February 2025; Revised: 19 April 2025; Accepted: 22 April 2025; Published: 15 May 2025 * Correspondence: catherine_meta@163.com Intelligent decision-making system for jewelry retail prediction and inventory management integrating CRM data Ting-Ting Han1* 1Jinqianzhi (Shenzhen) Software Technology Co., Ltd, Shenzhen 518019, China; catherine_meta@163.com (T.T.H.). Abstract: This study develops an intelligent decision support system that integrates Customer Relationship Management (CRM) data with inventory optimization to address the unique challenges of jewelry retail, characterized by high-value merchandise, emotional purchasing patterns, and seasonal demand fluctuations. A case-based empirical approach using semi-structured interviews and field observations (qualitative) alongside analysis of sales data, inventory records, and customer transactions (quantitative) is employed to evaluate the customer-centric inventory management system. Empirical evaluation in a mid-sized jewelry retail environment demonstrated significant performance improvements: a 23.5% increase in inventory turnover, 38.7% fewer stockout events, and a 14.6% higher customer satisfaction compared to control stores. The system enabled a "less inventory, better service" strategy, reducing total inventory by 12.3% while increasing product availability for high-value customers by 27.5%. The integration of CRM data with inventory management creates a transformative approach to retail operations, shifting from product-oriented to customer-oriented decision-making while simultaneously improving financial and service metrics. With a demonstrated ROI of 167% and an 18.6-month payback period, this study provides both a theoretical framework for blending customer data with inventory control decisions and a practical implementation guide for specialty retail environments. Keywords: CRM data integration, Customer-centric inventory, Intelligent decision-making system, Inventory optimization, Jewelry retail management, Predictive analytics, Retail digital transformation. 1. Introduction The retail industry, including jewellery, faces certain problems, such as the need for increasing customer experience customisation, high-value inventory, and seasonally fluctuating demand. Traditional inventory control models for jewellery retail typically rely on historical sales data and standard forecasting techniques that fail to account for the unique characteristics of the jewellery market, including emotional purchase decisions, significant seasonality, and the high value of individual items. Unfortunately, they tend to result in unused merchandise and inventory shortages, both of which can be very costly [1]. At the same time, the shift towards digitisation of almost all fields of economic activity is providing new opportunities for data capture and analysis within retail activities [2]. The combination of advanced technologies and the availability of big data gives the unprecedented ability to predict the behaviour of consumers and manage stock levels with incredibly precise accuracy [3]. A retail CRM has helped the sector tremendously by storing massive amounts of crucial information such as customer buying behaviour, preferences, demographics, and web activity [4]. This information is often neglected, particularly in specialty retail industries like jewellery [5]. Combining stock control systems with CRM data has proven to be very effective in making sales forecasts more precise, which improves inventory control. Still, this opportunity is limited because little academic research has been published regarding the gaps and problems of jewellery retail business in CRM-enabled decision making processes [6]. mailto:ahmed.raza@uos.edu.pk(A.R.C.) 1577 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate The fundamentals of jewellery retail differ from conventional trading because they involve emotional buying, high-value goods, seasonal variations in demand, and specialised customer service. These features necessitate the implementation of sophisticated sales forecasting and stock management approaches that traditional retailing methods might not be able to provide. Moreover, the jewellery sector has experienced significant expansion in eCommerce and omnichannel retailing, which complicates the forecasting and allocation of inventory through different channels [7]. This study attempts to construct an intelligent decision-making system capable of combining customer relationship management (CRM) data with sophisticated predictive analysis to enhance inventory control within the jewellery retail business. The design aims to bridge the gap between demand forecasting and inventory management in customer relationship management systems by using CRM data for inventory optimisation. The system’s approach to improving the prediction accuracy involves employing customer behaviour patterns, preferences, and historical purchase information while reducing inventory expenditures and increasing customer satisfaction. Jawad's research extends within the scope of the consumer behaviour and marketing domains as he tries to understand the need and significance of consumer preference information and its impacts on the decision-making processes within the deployment strategies of specialised retailing. From a theoretical standpoint, this research encourages the unification of customer value theory and resource-based theory in retail management. Practically, the proposed framework provides jewelry retailers with a competitive advantage by enabling better inventory turnover, reducing carrying costs, and increasing customer satisfaction through better product availability. 2. Literature Review 2.1. Research on Digital Transformation in Jewelry Retail The jewelry retail sector, as a specialized high-value retail domain, faces unique challenges and opportunities in digital transformation. With the rapid development of digital technologies, jewelry retailers are under pressure to transform traditional business models to meet evolving consumer demands.Hagberg, et al. [8] proposed an exploratory framework indicating that retail digitalization is not simply about transferring offline business online, but involves comprehensive transformation of retail exchanges, settings, actors, and offerings. In the jewelry retail sector, this transformation is particularly complex because the high-value nature of jewelry products and emotional purchase decision processes require specialized sales strategies and customer experience design. Shi, et al. [6] conducted case studies in the Chinese jewelry retail industry, examining the evolution of corporate innovation in the Online-to-Offline (O2O) model. Their findings depict how jewelry retailers have successfully merged online and offline channels through the mechanism of digital transformation, allowing for seamless customer experiences coupled with improved inventory management and supply chain effectiveness. Besides, Gouveia and São Mamede [9] emphasised the essential role of digital transformation for small and medium-sized retail businesses, claiming that the adoption of digital approaches can greatly enhance operational productivity, customer relations, and competitive advantage. A study conducted by Becker and Schmid [10] analysed both the strategy development and implementation stages of digital approaches in small and medium-sized firms to large ones, suggesting that the achievement of successful digital transformation is ultimately a matter of balancing strategy adherence with the necessary level of changeability. Regarding retail trade in jewellery, the incorporation of multi-channel information retrieval and analytical models to derive valuable customer data for personalisation of services is one of the major aspects of digital transformation [7]. As e-commerce and social media evolve, jewellery retailers are employing these tools to track consumer behaviour for better product design and sales strategies. Singh, et al. [11] noted that digital transformation not only changes a firm’s sales channels but also fundamentally transforms its organisational structure, business processes, and forms of value creation. 1578 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate 2.2. Analysis of Customer Relationship Management Systems CRM software plays a significant role in the retail business, particularly in the luxury jewellery market. Early research conducted by Swift [7] indicated that CRM technologies have the potential to enhance customer relations through customer information consolidation, behavioural analysis, and tailored service delivery, resulting in increased customer loyalty. The advent of big data technologies transformed modern CRM systems from basic customer data management software into intricate platforms capable of generating valuable analyses of customer behaviour. Phan and Vogel [4] proposed a conceptual model that merges customer relationship management and a business intelligence system that is designed for catalog and online retail businesses. This model captures the role of transactional data, customer profiles, and historical interactions as integral components of consumer behaviour. In the case of jewellery retail, this integration is of particular importance, as a variety of factors, such as personal preferences, special occasions, and emotional attachments tend to shape a consumer's purchasing behaviour. The study conducted by Sanguanpiyapan and Jasper [5] recognised the peculiarity of consumer behaviour patterns in jewellery shopping contexts, especially the necessity for consumer consciousness toward luxury goods in order to offer service experience that commensurates with such goods. Furthermore, Kumar, et al. [12] found that employing data mining and machine learning techniques in a retail setting has great possibilities of improving customer retention rates significantly. By analyzing data collected through CRM systems, retailers can predict customer behavior, identify churn risks, and take appropriate intervention measures. Jayasundara [13] further highlighted the importance of consumer analytics in marketing prediction, emphasizing that CRM data-driven analysis can achieve more precise market positioning and promotion strategy optimization. Although the value of CRM systems in retail has been widely recognized, Rothberg and Erickson [14] reminded researchers and practitioners of the need to distinguish between knowledge transfer and intelligence insights in big data systems. In jewelry retail environments, simply collecting customer data is insufficient; the key lies in transforming this data into actionable insights to guide decision- making processes for inventory management and sales strategies. 2.3. Research on Retail Prediction and Inventory Management Retail prediction and inventory management are core components of retail operations, directly impacting a company's financial performance and customer satisfaction. Kondo and Vicente [15] research emphasized the importance of coordinating customer demand and inventory management, proposing a comprehensive approach to enhance customer experience in retail. The study points out that effective inventory management is not merely about cost control but a key factor in providing superior customer experiences. Inventory management in jewellery retail comes with its own set of issues that include high-value items, volatile demand, and complex product life cycles. Shankar, et al. [1] noted that retail inventory control is being profoundly impacted by technology, with overstocking and stockout situations being mitigated through inventory level automation and predictive analytics. Bradlow, et al. [3] further developed the importance of big data and predictive analysis in retail, especially on demand estimation, price setting and personalisation through recommendation systems. Lo, et al. [16] set forth a framework for E-SCM multi-agent systems in the case of the fashion industry for inventory control at the supply chain level and pointed out the need for cooperation and information dissemination among supply chain constituents. Although the research concentrated on the fashion industry, its fundamental concepts are also relevant to jewellery retail, particularly in dealing with seasonal demand and managing inventory levels during product introductions. Grewal, et al. [2] examined retailing through the lens of emerging technologies and showed the efficacy of artificial intelligence and machine learning algorithms especially in improving demand forecasting precision, thus streamlining inventory decisions. Such technologies are useful in jewellery retailing because they enable retailers to understand intricate 1579 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate buying behaviour and seasonal changes, and consequently offer better instructions for inventory control. 2.4. Application of Big Data Analytics and Business Intelligence in Retail The implementation of big data analytics and business intelligence technologies in the retail sector is increasing exponentially, thereby giving organizations unparalleled insights. Charles, et al. [17] examined how data analytics and business intelligence allow retail organizations to extract key information that guides strategic decision-making and enhances operational efficiency. In the jewelry retail context, these technologies allow for deeper understanding of customer taste, forecast market trends, and maximize product offerings. Erevelles, et al. [18] carried out an extensive review of the revolutionary effects of big data consumer analytics on marketing strategies, citing that big data not only revolutionizes the way businesses collect and analyze consumer data but also radically transforms the relationships between businesses and consumers. In the jewelry retail context, the revolution is achieved through the use of increasingly personalized and contextually relevant marketing strategies, which are guided by extensive analysis of consumer purchasing behaviors, preferences, and trends. Ahmad, et al. [19] examined the role of business intelligence systems in contributing towards sustainable development in the textile and apparel sectors during the Industry 4.0 era. The study findings highlighted the critical role of data-driven decision-making in maximizing the use of resources and conserving the environment. Although the study's focus is different, its fundamental concepts are relevant to the jewelry retail sector, particularly in terms of inventory optimization and reducing resource wastage. Rothberg and Erickson [14] research reminds us that the value of big data systems lies not only in their ability to collect and store large amounts of data but more importantly in transforming this data into actionable intelligence insights. In jewelry retail, this means developing specialized analytical frameworks and algorithms to extract insights from CRM data relevant to inventory management and sales prediction. 2.5. Summary of Literature Review and Research Gaps Through the literature review on digital transformation in jewelry retail, customer relationship management systems, retail prediction and inventory management, and the application of big data analytics and business intelligence, several significant research gaps can be identified. First, while existing research emphasizes the importance of data-driven decision-making in retail, studies specifically targeting the jewelry retail sector, with its unique characteristics, are relatively limited. The high-value nature of jewelry products, emotional purchase decision processes, and seasonal demand patterns make general retail prediction models potentially challenging to apply effectively. Second, although the application of CRM systems in retail has been extensively studied, research on how to effectively integrate CRM data with inventory management systems to optimize jewelry retail operations remains insufficient. Existing research often treats customer relationship management and inventory management as relatively independent domains, lacking an integrated framework that organically combines both. Third, the integration of big data analytic tools into Business Intelligence (BI) systems in the retail sector is proving to be common practice, yet scant attention is paid to the specific characteristics and analytic requirements of the jewellery retail industry. The jewellery retail industry possesses data that is highly personalised, sparse, and has a long cycle, thereby necessitating the creation of specialised analytical techniques and predictive models tailored to these characteristics. Finally, the study of digital transformation tends to examine processes of organisational change and strategic modification from a bird’s eye view rather broad scope without any micro-implementation details for particular operational processes like sales forecasting and inventory control within the jewellery retail business. Due to this, jewellery retailers find it challenging to put digital strategies into specific operational refinement actions at the operational level. 1580 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate This research intends to target these gaps by proposing an intelligent decision support system for jewellery retail sales forecasting and inventory control that collects and processes CRM data, thus offering a decision support system for the jewellery retail business. The system will integrate CRM and inventory management, with the goal of utilising customer data to enhance demand predictions and optimise stock levels, thereby providing different angles and ways for new thinking and approaches for both theoretical researchers and practical applications. 3. Theoretical Framework and Intelligent Decision-Making System Design 3.1. Theoretical Framework The present research uses an eclectic approach that integrates resource-based theory and customer value theory, emphasising operational efficiency and customer orientation as essential factors of competitiveness in the jewellery retailing market. A firm can maintain a competitive edge in the long run through the effective utilisation of its resources and competencies, which is a primary assumption of the resource-based theory [20]. In the case of jewellery retailing, inventory is regarded as a prime asset that requires sophisticated management capability. In parallel, customer value theory suggests that loyal customers and profits are the results of effective value-added processes [21]. This study posits that effective inventory management should achieve a certain level of operational efficiency while also meeting customer satisfaction, therefore merging these theoretical perspectives. The model depicted in Figure 1 combines three concepts within a single framework: integration of information, prediction, and decision making. In the case of information integration, the retail industry is known to contain a variety of data sources, whereas CRM data is identified as one of the primary sources of customer information. This approach is related to the theory of knowledge management which argues that there is a need to collect and utilise organisational knowledge and information [22]. In contrast, the predictive analytics aspect, which stems from decision theory as well as blending traditional statistics with more modern approaches, more commonly referred to as data science, aims at producing accurate demand estimates from the available information [23]. Finally, the application of optimisation in decision making is based on the use of operations research methods which focus on how the forecasts can be applied in a manner that results in the best possible decisions regarding inventory control [24]. Customer Value Theory Decision Theory Resource-Based Theory Knowledge Management Theory Data Integration Dimension Predictive Analytics Dimension Decision Optimization Dimension Customer Responsiveness Inventory Optimization Financial Control Theoretical Foundations Framework Dimensions Implementation Outcomes Figure 1. Theoretical Framework for Intelligent Decision-Making in Jewelry Retail. 1581 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate A distinctive feature of this framework is its cyclical and adaptive nature, which reflects the dynamic environment of jewelry retail. Unlike linear models, this framework acknowledges feedback loops between decision outcomes and subsequent data collection and analysis processes. This approach aligns with dynamic capabilities theory Teece, et al. [25] emphasizing organizational adaptation to changing market conditions through continuous reconfiguration of resources and capabilities. The framework also incorporates contingency perspectives, recognizing that the optimal approach to jewelry retail inventory management is context-dependent, providing flexibility to adapt the intelligent decision- making system to diverse retail environments while maintaining theoretical coherence. 3.2. Research Methodology and Design This research employs a mixed-methods approach combining system development methodology with empirical validation through case study research. The methodology follows a phased progression from conceptual development to practical implementation and evaluation, ensuring both theoretical rigor and practical relevance. The initial phase involves comprehensive literature analysis and expert interviews to refine the theoretical framework and system requirements. Subsequently, the system development phase employs agile methodologies to iteratively build and refine the intelligent decision- making system. The research design incorporates both qualitative and quantitative elements to address the multifaceted nature of the research objectives. Qualitatively, in-depth interviews and observational techniques are employed to understand existing jewelry retail inventory management practices and identify pain points. Quantitatively, historical sales and inventory data from participating jewelry retailers are analyzed to establish baseline performance metrics and validate the predictive accuracy of the developed system. Data collection strategies capture the full spectrum of relevant variables, including customer purchase history, demographic information, browsing behavior, inventory turnover rates, stockout incidents, and financial performance indicators. The validation strategy employs a multi-level approach, evaluating technical performance metrics, operational workflow integration, and strategic business impacts. 3.3. Overall System Architecture The intelligent decision-making system is designed as a modular, scalable architecture that facilitates seamless integration with existing retail management systems while providing advanced analytical capabilities. As shown in Figure 2, the system architecture consists of four primary layers: the data layer, the integration layer, the analytics layer, and the decision support layer. Each layer serves specific functions while maintaining clear interfaces with adjacent layers, promoting system flexibility and adaptability. 1582 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate Web Analytics & Social Media Inventory Management Data POS Transaction Data CRM Data Sources Demand Prediction Customer Profiling Seasonal Pattern Optimization Sentiment Analysis Performance Feedback Product Mix Recommendation Inventory Optimization Pricing Recommendation User Interface Data Extraction. Transformation & Loading Data Quality & Governance Master Data Management Data Layer Integration Layer Analytics Layer Decision support Layer Figure 2. Intelligent Decision-Making System Architecture for Jewelry Retail. The data layer, as shown in Figure 2, is made up of different data sources such as CRM systems, point of sale (POS) systems, inventory systems, and web data. The integration layer solves data fragmentation heterogeneity with ETL processes, data cleansing, and master data management. The analytics layer is the most sophisticated part of the system whose purpose is to analyse the data and provide advanced business intelligence results which can be acted upon. In the decision support layer, recommendation systems use the resulting analytical models to provide practical suggestions for inventory control by optimising many conflicting goals. The system and jewellery retail managers interface combine their functions in a single, user-friendly interface that allows the visualisation of the results of the analysis and decision suggestions. Furthermore, the system improves automatically by observing user behaviour and the results of their decisions. These actions are sent back into the analysis layer to improve the predictive models and optimisation algorithms. 3.4. CRM Data Integration Module The CRM Data Integration Module is an essential part of the intelligent decision-making system which uses customer-focused data for inventory management. This module converts customer relationships data, which is usually used for marketing and sales, into formats that can effectively respond to inventory requirements. The module processes diverse customer-related data, including transactional records, customer profile information, behavioural data, and communication history. 1583 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate This module performs some degree of data preprocessing to improve quality and consistency, data transformation for customer data into specific meaning metrics and patterns, and data enrichment for further contextual external information to improve analytical value. The output of integration consists of customer-derived metrics, customer influence parameters, and customer preference models which serve as a flexible base for inventory investment decisions. A feature that distinguishes this module is its omnidirectional information flow with the analytics and decision support systems which enables adaptive revision of customer models with market and customer behaviour change. 3.5. Sales Prediction and Inventory Management Modules The Sales Prediction and Inventory Management Modules form the analytical core of the intelligent decision-making system, transforming integrated data into actionable inventory recommendations. These modules address the dual challenges of forecasting volatile jewelry demand with high accuracy and translating these forecasts into optimal inventory decisions that balance financial and service level objectives. As illustrated in Figure 3, these interrelated modules have a sequential yet feedback-driven relationship. Historical Time Series Analysis Customer Segment Contribution External Factors Prediction Model Fusion Techniques Forecast Accuracy Safety Stock Optimization Multi-location Planning Automated Replenishment Inventory Mix Optimization Scenario Analysis Inventory Turnover Service Level Sales Prediction Module Inventory Management Module Demand Forecasts Model Refinement Process Optimization Performance Monitoring Figure 3. Integration of Sales Prediction and Inventory Management Modules. As shown in Figure 3, the Sales Forecasting Module takes a holistic approach to forecasting jewelry demand based on multiple product categories, channels of distribution, and temporal considerations. The module combines a review of past time series data, the modeling of the contributions of different customer segments, and the forecasting of external variables using model fusion techniques. The approach is overarching because it understands that jewellery consumption is a direct result of 1584 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate numerous factors such as previous fashion styles, shifts in the consumer population, and even subordinate factors like the current economic state and times of year. The Inventory Management Module converts the expected sales into operational inventory plans by overshooting plan safety stock, permitting consolidated planning across multiple locations, and implementing automatic reorder systems. Order execution methods are defined to provide the required inventory levels, disburse stock to retail stores, and devise efficient means of ordering stock. Furthermore, the integrated framework for inventory management includes maximisation of inventory mix, scenario evaluation, and logic for minimising the reordering of stock-in-trade while endeavouring to provide a structure that incorporates product variety, risk analysis, and implementation feasibility constraints. The performance measurement process constitutes an important feedback that enables change within various modules and at the same time validates estimates, inventory turn rates, and qualitative metrics pertaining to service levels. Feedback loops facilitate adaptive learning by allowing model parameters, optimisation boundaries, and business rules to be constantly adjusted based on the actual outcomes achieved. 3.6. Integration and Implementation Procedures for Systems The successful deployment of an intelligent decision support system necessitates careful integration with the existing retail infrastructure, which must be accompanied by clear deployment plans. This requires establishing means for communication exchange, determining integration locations, and providing assimilation into operational activities. This process is gradual, starting with the involvement of stakeholders to assist in achieving the organisational objectives, and then moving to pilot deployments at selected strategically beneficial sites to evaluate how the system performs in actual work settings prior to full deployment within the organisation. Change management is one of the most important steps in realising technological innovations, and organisational design facilitates. To ensure an appropriate degree of longevity and value capture, numerous training, performance review, and continual development processes are set in place. Furthermore, the implementation phase includes capturing user experience and performance information so that these processes can be refined and adjusted to the evolving business context over time. 4. Empirical Research and Case Analysis 4.1. Research Design and Data Collection This research employs a case-based empirical approach to evaluate the implementation effectiveness of an intelligent decision-making system for jewelry retail that integrates CRM data with inventory management. The research methodology combines qualitative and quantitative methods to provide a comprehensive assessment of both technical implementation and business impact. Qualitatively, semi- structured interviews and field observations were conducted to understand technical challenges and solutions during system implementation; quantitatively, sales data, inventory records, and customer transaction data were analyzed to assess prediction accuracy and inventory management efficiency. The data collection process was structured in three phases to ensure effective comparison before and after system implementation. In the baseline phase, we collected 12 months of sales data, inventory records, and customer segmentation data from the case company's existing systems. During the system development and testing phase, we gathered technical data on algorithm performance, prediction accuracy, and system response times. In the post-implementation phase, we tracked changes in key performance indicators to evaluate system effectiveness. Table 1 summarizes the data sources, collection methods, and sample sizes used in this research. The research protocol involving human participants was reviewed and approved by the institutional review board, and all participants provided informed consent prior to their participation in the study. 1585 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate Table 1. Data Sources and Collection Methods. Data Category Source Collection Method Sample Size Purpose Transaction Data POS System API Extraction 4,873 transaction records Model training, sales pattern analysis Inventory Data Inventory System System Export 12 months, 486 SKUs Inventory turnover analysis, optimization validation Customer Data CRM System Structured Query 1,247 customer profiles Customer segmentation, behavior analysis System Performance Decision System System Logs 3 months of operation logs Technical performance evaluation, algorithm optimization User Feedback System Operators Structured Questionnaire 23 system users Usability assessment, interface optimization The research specifically focuses on technological innovation and implementation methods, aligning with the technical orientation of applied science journals. By documenting system architecture design, algorithm selection and optimization processes, data integration methods, and user interface design, this research provides a replicable technical framework for similar system development in specialized retail environments. The data analysis employs a mixed-method approach, with quantitative analysis primarily utilizing descriptive statistics and forecast accuracy metrics, while qualitative analysis employs thematic coding to identify key technical challenges and solution strategies in the implementation process. 4.2. Case Company Background Company G operates within the jewellery retailing field, which is wealth inventory intensive, has considerable seasonal demand variations, and has distinct customer service requirements. Since 2023, the company has been under greater difficulty due to fragmentation brought about by digital transformation and omnichannel retailing. Customers have actively changed their shopping habits to prefer purchasing electronically, where the traditional distinction between physical retail and online shopping is increasingly vague, creating new challenges for traditional inventory management systems that predominantly depend on historical sales data and subjective judgement decision making approaches. Before the system implementation, Company G had the classic “information silos” servicing their business information system without any integration and sophisticated level of analytics. The ERP and POS systems, which were doing some automation of the business processes, had some rudimentary associations with the CRM system that was installed in 2021. Lecture incrementation during course partitioning prevented the use of sales analysis from guiding stock decisions, which made stock allocation not only less than optimal, but unaligned with what the customers wanted and what was actually purchased. Further, the company had very limited stock of analytical tools, and decisions on stock levels were predominantly based on experience and simple analysis of historical data, leading them to approximately 67.3% accuracy of forecasts, and 2.2 stock turnover per year which is very low compared to the industry average of 3.4. To solve these problems, in late 2023 Company G made the choice to create and put into use an intelligent decision-making system amalgamating CRM and inventory data. These systems were aimed at improving the accuracy of sales forecasts, optimising inventory levels, reducing stockout rates, and elevating customer satisfaction. Company G’s management understood that customer data integration together with inventory management could facilitate the shift from product driven inventory management to customer driven inventory management, thereby gaining a competitive edge with intelligent decision making. The system was fully operationalised by September 2024 and its development commenced in January 2024 which provided relevant data for the study. 1586 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate 4.3. System Implementation Process The agile approach was adopted for the smart decision-making system’s design and implementation, spanning approximately nine months, from January 2024 to September 2024. The entire implementation process was broken down into four core components: requirements gathering and design, system implementation, verification and validation, and system installation. Each component captured a set of specific technical objectives alongside deliverables which made it easier to transition from conception to implementation in a systematic manner. The project team conducted the detailed business analysis to define the key functional requirements that were needed in the system during the requirements gathering and design stage that took place from January to February 2024. These requirements included modules to integrate Customer Relationship Management data, sales data forecasting, and inventory levels optimisation. A microservice architecture design was utilised which decomposed system functions into independently deployable service modules, thus improving system flexibility and scalability. This architecture was critical in resolving the challenging data flows reporting customer insights together with inventory control problems by providing a complete data pipeline from analysing customer behaviour to optimising inventory levels. The technical design integrated functional requirements, performance, and offered data models with algorithm selection criteria, integration definitions, and graphical user interface design rules that would be used to mould subsequent development work. March to June 2024 represented the first phase of the implementation cycle, the period during which coding and unit testing for key functional modules were finalised. The customer relationship management (CRM) data integration module employed the extract, transform, load (ETL) process to merge and cleanse customer data from various databases, thus addressing major problems concerning data quality and structural inconsistencies. Furthermore, the sales forecasting module integrated time series analysis and machine learning to create a hybrid forecasting model tailored to the jewellery retail business. The model utilised classical ARIMA methods to capture the seasonal effects with gradient boosting techniques that took customer segment attributes into consideration for jewellery’s highly volatile demand. The inventory optimisation module relied on operations research concepts and formulated a multi-objective inventory optimisation problem that sought to minimise inventory levels and maximise service levels across customer segments. The system frontend used responsive design concepts, providing easy-to-use data visualization and decision support interfaces that encouraged user interaction and supported effective decision-making. Testing and validation (July-August 2024) ensured system functional integrity and performance reliability through multi-level testing approaches. Unit testing verified the correctness of individual functional modules, integration testing examined data flow and interface compatibility between modules, and system testing evaluated overall functionality and performance under realistic operational conditions. Notably, the testing phase employed historical data backtesting to validate prediction model effectiveness by comparing system predictions with actual historical data. Performance testing revealed that the hybrid prediction model achieved a Mean Absolute Percentage Error (MAPE) of 8.3%, significantly outperforming the company's previous forecasting approach with a MAPE of 23.4%. User acceptance testing gathered feedback from key stakeholders, resulting in interface refinements and workflow optimizations that enhanced usability and user experience. The deployment phase (August-September 2024) implemented a staged rollout strategy, beginning with a pilot in five stores to validate system performance in actual business environments before gradually expanding to all stores. Throughout deployment, the implementation team focused on user training and technical support, developing training courses, operation guides, and online help resources to enhance users' understanding and application capabilities. A problem tracking and feedback mechanism was established to quickly address technical issues and functional requirements encountered during deployment, ensuring smooth transition and effective knowledge transfer. By mid-September 2024, the system had been successfully implemented across all company-owned stores, with key users demonstrating proficiency in system operation and decision making based on system recommendations. 1587 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate Table 2 summarizes the major technical challenges encountered during system development and the innovative solutions implemented, reflecting the specific system development requirements and technical difficulties in the jewelry retail industry. Table 2. Major Technical Challenges and Solutions. Technical Challenge Description Solution Innovation Heterogeneous System Integration CRM and ERP systems using different data formats and models Middleware data transformation services with semantic mapping technology Ontology-based semantic integration method improving mapping accuracy Jewelry Demand Prediction Complexity High volatility jewelry demand influenced by multiple factors Hybrid prediction model integrating time series analysis with customer segmentation features Customer behavior-weighted LSTM neural network model enhancing prediction accuracy Multi-objective Inventory Optimization Balancing inventory costs, service levels, and product diversity Multi-objective optimization algorithm incorporating customer value dimension Customer value-tiered differential inventory strategy optimizing resource allocation System Performance System response delays due to large data processing and complex algorithm computation Data preprocessing, result caching, and algorithm optimization Gradient computation parallelization and incremental update strategy improving algorithm efficiency User Interface and Decision Support Complex analytical results difficult to present intuitively Interactive visualization interface with decision recommendations and explanations Explainable AI-based recommendation system enhancing user understanding and trust The system architecture implemented a modern technology stack with a layered design that ensured clear data flow and modular functionality. The frontend layer provided user interaction interfaces and data visualization capabilities; the backend layer implemented core business logic, including prediction algorithms and optimization models; the data layer managed data acquisition, processing, and storage; while the data source layer connected to the company's existing CRM, ERP, and POS systems. This architectural design not only met current business requirements but also demonstrated excellent scalability to accommodate future business changes and technological developments. 4.4. Implementation Effect Evaluation After the system was implemented, a comprehensive evaluation was conducted to determine its effectiveness with regards to technical performance, business impact, and user experience factors. This evaluation used controlled experimental designs and before-and-after comparative methods to quantitatively and qualitatively analyze the concrete advantages of the system using measurable parameters and user feedback. The technical performance evaluation focused on the functional completeness of the system, operational dependability, and computational efficiency, using log analysis and performance testing techniques. The results proved that the system achieved or approached design targets in most key technical parameters. In particular, the accuracy of predictions, measured by the Mean Absolute Percentage Error (MAPE), reached 12.4%, exceeding the target level of below 15%, but still showing room for improvement. System availability reached 98.6%, guaranteeing business continuity, while concurrent user support capacity of 71 users satisfied the company's operational needs. Business impact evaluation analyzed the system's effect on inventory management and sales performance using a controlled experiment methodology. Stores using the system (experimental group) were compared with stores not using the system (control group) to assess the net effect of system implementation. Figure 4 visually demonstrates the comparative performance improvements between the experimental and control groups across key business metrics. 1588 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate Figure 4. Comparative Business Impact Between Experimental and Control Groups. As shown in Figure 4, the experimental group demonstrated notably higher improvements across all metrics. Inventory turnover increased by 23.5% in stores using the system compared to 11.8% in the control group. Stockout rates decreased by 38.7% versus 17.5%, while obsolete inventory ratio reduced by 31.4% versus 14.9%. Customer satisfaction scores improved by 14.6% in the experimental group compared to only 5.2% in the control group. Gross margin improvement reached 8.9% versus 3.1% in the control group. These differences demonstrated that system implementation, rather than external factors, drove business improvement. The system improved inventory structure, achieving more precise inventory investment. Analysis revealed that post-implementation inventory configuration better matched customer demand, with product availability for high-value customer segments increasing by 27.5% while overall inventory levels decreased by 12.3%. This "less inventory, better service" outcome validated the effectiveness of the CRM-integrated intelligent decision system in optimizing resource allocation. Economic analysis calculated the system's return on investment at 167%, with a payback period of approximately 18.6 months, demonstrating financial justification for the implementation. User experience evaluation through questionnaires and interviews gathered feedback from system users. Overall system satisfaction reached 3.8 on a 5-point scale, with 71.3% of users reporting improved decision efficiency and 65.8% reporting enhanced decision confidence. Users also provided improvement suggestions, including enhancing system explainability, simplifying operational processes, and providing more diverse analytical views, providing valuable direction for continuous system optimization. The implementation process encountered several significant challenges. In the initial phase, data quality issues caused prediction accuracy instability, particularly for new product categories where the average error rate reached 26.3%. Approximately 18% of high-end jewelry products were difficult to predict accurately due to sparse historical data, requiring additional manual adjustments. During the integration process, about 7% of customer data experienced matching errors, necessitating the development of additional data cleaning processes. User acceptance also showed polarization, with 1589 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate approximately 22% of senior staff adopting the new system at a notably slower rate than younger employees, requiring additional training resources. The technical innovations and solutions developed during system implementation also generated knowledge assets, including a hybrid prediction model algorithm incorporating customer behavior features and a customer value-based differential inventory optimization method. These technical innovations not only solved specific business problems but also demonstrated broader application potential that could extend to other retail domains, particularly high-value product retail industries with similar characteristics. Some technical limitations and improvement opportunities were identified. The current prediction model shows limited adaptability when handling extreme events such as sudden product trends. The system's heavy dependence on high-quality data means prediction accuracy suffers when input data quality is poor. Integration with external data sources such as social media sentiment analysis remains incomplete, limiting proactive understanding of consumer trends. These limitations provide clear direction for future system upgrades. The comprehensive evaluation results confirm that the CRM-integrated intelligent decision-making system for jewelry retail prediction and inventory management successfully achieved its intended objectives, demonstrating positive outcomes in both technical implementation and business application. The system not only improved prediction accuracy and inventory management efficiency but also facilitated a business transformation from product-oriented to customer-oriented operations, providing jewelry retailers with an effective tool for establishing competitive advantage in the digital environment. 5. Discussion 5.1. Main Research Findings The intelligent decision support system for jewellery retail within the CRM datatssystem captures an integration feature as it’s implemented and evaluated tells us much about the innovation of specialised retail trade. The use of the system in Company G proves that integrating customer relationship data with stock control systems serves to improve operational results considerably in the jewellery retail business. The accuracy of forecasting with a hybrid prediction model that combines time series analysis and features of customer behaviour was 87.6% (MAPE: 12.4%). This improvement is astounding compared to the results from conventional forecasting approaches. This confirms the core assertion that demand forecasting for sophisticated retailing is much better with the presence of customer-centric data, especially where the purchasing decision is greatly impacted by personal emotions and preferences. Without differentiating customer segments, the traditional inventory models are proven to be inefficient. Fulfilling the particular customer’s needs by using their purchasing intelligence will lead to what he termed as ‘less inventory, better service’. Inventory optimisation that is guided by customer insights resulted in the overall inventory being reduced by 12.3%, whilst product availability for high- value customer segments increased by 27.5% for specific customers. There is sufficient evidence to show that the system can optimise inventory stock models. The 23.5 percent increase in inventory turnover and the 38.7 percent decrease in the stockout rates in the experimental group supports the efficacy of the system. It appears that these changes are due to the system’s enhanced capability of analysing customer purchase patterns and preferences with greater precision. 5.2. Theoretical and Practical Implications This research greatly contributes to the theoretical and practical foundations of retail operations management and information systems integration. Theoretically, it builds on the interaction between customer value theory and resource-based theory by demonstrating how customer knowledge can enhance the effectiveness of resource allocation processes. This research explores how a customer- centric approach can be incorporated into inventory optimisation algorithms, thus extending the application of resource-based theory for the customer-oriented decision-making processes. The provided 1590 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate evidence demonstrates how fulfilling resource allocation strategies, which are designed with the diversity of customer demands in mind, can achieve customer and operational satisfaction simultaneously. This finding is in opposition to the common belief which assumes that the two objectives cannot be accomplished at the same time. This framework allows the user to convert customer relationship management information into operational knowledge, and hence adds to the theoretical base in knowledge management. The system is based on a new approach to semantic integration, which solves an essential problem of domain knowledge when such knowledge is encapsulated in functional silos. It establishes a form of 'operational customer intelligence' that intervenes in inventory decision-making processes by developing semantic patterns between customer behaviours and inventory parameters. This transformation of knowledge is an example of the application of functional integration, which has implications beyond the retail context. This research advances the conception of a pragmatic technical model capable of merging customer relationship management (CRM) and inventory management (IM) in such a way as to permit retailers to shift from independent information systems to integrated decision-making systems. The microservices architecture and the semantic integration approach demonstrated in this work can act as an architectural pattern for other integration projects of this nature. Furthermore, the exogenous hybrid predictive model, which combines the attributes of customer behaviour, offers a practical way of enhancing forecasting accuracy in ever-changing retail markets and indicates that retailers must focus on analytics of customer behaviour in order to demand more precise forecasts. This system's inventory management and control strategies, which classify customers by the value they add, balance the financial burden versus service level requirements for high-end retailing remarkably well. The justification in econometric terms is clear with advanced ROI decision-making resulting in a 167% return and an 18.6-month payback period. Moreover, the results derived from the implementation project substantiate the claims regarding user education and effective organisational change as vital in the system's acceptance as reflected in user ratings satisfaction of 3.8 over 5 and 65.8% of users having more confidence in decision-making. Both, in addition to functional outcomes, are regarded as very important influences alongside technical performance measures. 5.3. Research Limitations and Future Research Directions Notwithstanding the enormous input, this research suffers from several gaps that may be pursued in subsequent studies. First, the case study method offers profound understanding of system implementation; however, it does not allow for broader application in different retail settings. Results are based on one firm in the Chinese jewellery market and the system’s efficacy is dependent on the market’s environment or retail segments. This is especially true for markets in emerging economies. Further studies are needed to assess the system’s usability in various retail settings, including multicultural settings and other product categories. Second, the current system has a low degree of flexibility for coping with extreme events, such as sudden consumer fads or market chaos. The prediction model worked well with average accuracy for normal market situations, but there is likely to be greater need for change to accommodate unanticipated market changes. This was particularly true for new product categories, where the average error in all such categories was recorded at 26.3%. Additionally, around 18% of luxury jewelry products posed substantial challenges to accurate forecasting due to thin-slicing effects. Future research efforts can be directed toward improving advanced methods for anomaly detection and designing adaptive learning algorithms that can efficiently handle market irregularities while being highly predictive. The existing implementation is significantly short on integrating external sources of data, such as social media sentiment analysis and macroeconomic signals. Future work should focus on the formulation of sophisticated data integration methods that are capable of identifying real-time consumer sentiment and external market data to improve the system's predictability. Another key area for future research includes investigating the impact of blockchain technology on the reliability and traceability of inventory management systems in the jewellery industry. 1591 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate In summary, while the superficial gains in operational effectiveness are clear, the interaction of the system with the elements of customer loyalty and lifetime value constitutes a more complex problem that calls for longitudinal analysis. Based on this assumption, future research should investigate how better inventory synchronization matched to customer needs affects customer retention and total financial performance over long periods. A study of this question would shed light on how smart decision-making systems in specialty retail environments create significant strategic benefits over time that are not only positive but also highly informative. 6. Conclusion This research has successfully developed and implemented an intelligent decision-making system for jewelry retail prediction and inventory management that integrates CRM data. The system addresses the unique challenges of the jewelry retail sector by establishing a customer-centric approach to inventory optimization, effectively bridging the gap between customer relationship management and inventory control functions. Through empirical evaluation in Company G, we have demonstrated that this integrated approach significantly improves key performance metrics including inventory turnover (23.5% increase), stockout rate reduction (38.7%), and customer satisfaction (14.6% improvement). The system's core innovations—a hybrid prediction model incorporating customer behavior features and a customer value-based differential inventory optimization method—represent significant technical contributions to specialized retail management. By achieving the paradoxical outcome of "less inventory, better service," the system demonstrates that customer insights can fundamentally transform resource allocation efficiency in high-value retail environments. The system reduced overall inventory levels by 12.3% while simultaneously increasing product availability for high-value customer segments by 27.5%. From a theoretical perspective, this research advances the integration of resource-based theory and customer value theory, establishing a framework for cross-functional knowledge utilization that transforms customer data into operational intelligence. From a practical standpoint, the system's architecture and implementation methodology provide a replicable blueprint for retailers seeking digital transformation, with a demonstrated ROI of 167% and payback period of 18.6 months justifying the investment. While the current implementation has demonstrated substantial business value, future research should explore enhanced adaptability for extreme market events, particularly for addressing the 26.3% error rate found in new product categories and the prediction challenges for the 18% of high-end jewelry products with sparse historical data. Additional areas for improvement include integration of external data sources including social media sentiment, and longitudinal studies examining long-term impacts on customer loyalty. As digital transformation continues to reshape retail landscapes, the intelligent integration of customer insights with operational decisions will become increasingly critical for establishing sustainable competitive advantage in specialized retail environments. Transparency: The author confirms that the manuscript is an honest, accurate, and transparent account of the study; that no vital features of the study have been omitted; and that any discrepancies from the study as planned have been explained. This study followed all ethical practices during writing. Copyright: © 2025 by the author. This open-access article is distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/). References https://creativecommons.org/licenses/by/4.0/ 1592 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 5: 1576-1592, 2025 DOI: 10.55214/25768484.v9i5.7235 © 2025 by the author; licensee Learning Gate [1] V. Shankar et al., "How technology is changing retail," Journal of Retailing, vol. 97, no. 1, pp. 13-27, 2021. https://doi.org/10.1016/j.jretai.2020.10.006 [2] D. Grewal, D. K. Gauri, G. Das, J. Agarwal, and M. T. Spence, "Retailing and emergent technologies," Journal of Business Research, vol. 134, pp. 198-202, 2021. https://doi.org/10.1016/j.jbusres.2021.05.004 [3] E. T. Bradlow, M. Gangwar, P. Kopalle, and S. Voleti, "The role of big data and predictive analytics in retailing," Journal of Retailing, vol. 93, no. 1, pp. 79-95, 2017. https://doi.org/10.1016/j.jretai.2016.12.004 [4] D. D. Phan and D. R. Vogel, "A model of customer relationship management and business intelligence systems for catalogue and online retailers," Information & Management, vol. 47, no. 2, pp. 69-77, 2010. https://doi.org/10.1016/j.im.2009.09.001 [5] T. Sanguanpiyapan and C. Jasper, "Consumer insights into luxury goods: Why they shop where they do in a jewelry shopping setting," Journal of Retailing and Consumer Services, vol. 17, no. 2, pp. 152-160, 2010. https://doi.org/10.1016/j.jretconser.2009.12.001 [6] M. Shi, C. Zhang, and C.-L. Chen, "The Evolution of Corporate Innovation in the O2O Model—Case Studies in the Chinese Jewelry Retail Sector," Sustainability, vol. 15, no. 17, p. 13017, 2023. https://doi.org/10.3390/su151713017 [7] R. S. Swift, Accelerating customer relationships: Using CRM and relationship technologies. Upper Saddle River: Prentice Hall Professional, 2001. [8] J. Hagberg, M. Sundstrom, and N. Egels-Zandén, "The digitalization of retailing: An exploratory framework," International Journal of Retail & Distribution Management, vol. 44, no. 7, pp. 694-712, 2016. https://doi.org/10.1108/ijrdm-09-2015-0140 [9] F. D. Gouveia and H. São Mamede, "Digital transformation for SMES in the retail industry," Procedia Computer Science, vol. 204, pp. 671-681, 2022. https://doi.org/10.1016/j.procs.2022.08.081 [10] W. Becker and O. Schmid, "The right digital strategy for your business: An empirical analysis of the design and implementation of digital strategies in SMEs and LSEs," Business Research, vol. 13, no. 3, pp. 985-1005, 2020. https://doi.org/10.1007/s40685-020-00124-y [11] S. Singh, M. Sharma, and S. Dhir, "Modeling the effects of digital transformation in Indian manufacturing industry," Technology in Society, vol. 67, p. 101763, 2021. https://doi.org/10.1007/s40685-020-00124-y [12] M. R. Kumar, J. Venkatesh, and A. M. Z. Rahman, "Data mining and machine learning in retail business: Developing efficiencies for better customer retention," Journal of Ambient Intelligence and Humanized Computing, pp. 1-13, 2021. https://doi.org/10.1007/s12652-020-02711-7 [13] C. C. Jayasundara, "Prediction of marketing by the consumer analytics. In Big Data Analytics." Boca Raton: Auerbach Publications, 2022, pp. 77-118. [14] H. N. Rothberg and G. S. Erickson, "Big data systems: knowledge transfer or intelligence insights?," Journal of Knowledge Management, vol. 21, no. 1, pp. 92-112, 2017. https://doi.org/10.1108/jkm-07-2015-0300 [15] K. Kondo and Â. J. B. Vicente, "The coordination imperative: A comprehensive approach to align customer demand and inventory management for superior customer experience in retail," Doctoral Dissertation, Massachusetts Institute of Technology, 2023. [16] W.-S. Lo, T.-P. Hong, and R. Jeng, "A framework of E-SCM multi-agent systems in the fashion industry," International Journal of Production Economics, vol. 114, no. 2, pp. 594-614, 2008. https://doi.org/10.1016/j.ijpe.2007.09.010 [17] V. Charles, P. Garg, N. Gupta, and M. Agarwal, "Data analytics and business intelligence," Data Analytics and Business Intelligence, 2023. https://doi.org/10.1201/9781003189640 [18] S. Erevelles, N. Fukawa, and L. Swayne, "Big Data consumer analytics and the transformation of marketing," Journal of Business Research, vol. 69, no. 2, pp. 897-904, 2016. https://doi.org/10.1016/j.jbusres.2015.07.001 [19] S. Ahmad, S. Miskon, R. Alabdan, and I. Tlili, "Towards sustainable textile and apparel industry: Exploring the role of business intelligence systems in the era of industry 4.0," Sustainability, vol. 12, no. 7, p. 2632, 2020. https://doi.org/10.3390/su12072632 [20] J. Barney, "Firm resources and sustained competitive advantage," Journal of Management, vol. 17, no. 1, pp. 99-120, 1991. https://doi.org/10.1177/014920639101700108 [21] R. B. Woodruff, "Customer value: The next source for competitive advantage," Journal of the Academy of Marketing Science, vol. 25, pp. 139-153, 1997. https://doi.org/10.1007/bf02894350 [22] M. Alavi and D. E. Leidner, "Knowledge management and knowledge management systems: Conceptual foundations and research issues," MIS Quarterly, pp. 107-136, 2001. https://doi.org/10.2307/3250961 [23] T. Hastie, R. Tibshirani, J. H. Friedman, and J. H. Friedman, "The elements of statistical learning: data mining, inference, and prediction." New York: Springer, 2009, pp. 1-758. [24] F. S. Hillier and G. J. Lieberman, Introduction to operations research. New York: McGraw-Hill, 2015. [25] D. J. Teece, G. Pisano, and A. Shuen, "Dynamic capabilities and strategic management," Strategic Management Journal, vol. 18, no. 7, pp. 509-533, 1997. https://doi.org/10.1002/(sici)1097-0266(199708)18:7%3C509::aid- smj882%3E3.0.co;2-z https://doi.org/10.1016/j.jretai.2020.10.006 https://doi.org/10.1016/j.jbusres.2021.05.004 https://doi.org/10.1016/j.jretai.2016.12.004 https://doi.org/10.1016/j.im.2009.09.001 https://doi.org/10.1016/j.jretconser.2009.12.001 https://doi.org/10.3390/su151713017 https://doi.org/10.1108/ijrdm-09-2015-0140 https://doi.org/10.1016/j.procs.2022.08.081 https://doi.org/10.1007/s40685-020-00124-y https://doi.org/10.1007/s40685-020-00124-y https://doi.org/10.1007/s12652-020-02711-7 https://doi.org/10.1108/jkm-07-2015-0300 https://doi.org/10.1016/j.ijpe.2007.09.010 https://doi.org/10.1201/9781003189640 https://doi.org/10.1016/j.jbusres.2015.07.001 https://doi.org/10.3390/su12072632 https://doi.org/10.1177/014920639101700108 https://doi.org/10.1007/bf02894350 https://doi.org/10.2307/3250961