Sentiment Analysis

SENTIMENT ANALYSIS IN DETERMINING CUSTOMERS PREFERENCES IN FASHION: A CROSS-NATIONAL PERSPECTIVE Adishi Lauretta 1 & Ugochukwu Matthew O. 2* 1 Fashion Department, Hussaini Adamu Federal Polytechnic Kazaure, Jigawa. 2 Computer Science Department, Hussaini Adamu Federal Polytechnic Kazaure, Jigawa. *[email protected] Abstract In recent years, the traditional clothing market has been threatened by many other new shopping channels. However, few researches on fashion marketing have been found to give systematic suggestions for traditional fashion stores in order to improve their marketing strategies to compete with other strong competitors. Therefore, this paper was aimed at understanding consumer’s purchase intention in traditional clothing purchase channel, in order to put forward corresponding marketing methods for traditional fashion stores to improve their strategies. Based on five purchases decision-making stages of EKB model, this study employed structured questionnaires and hypothesis testing with a sample of consumers from three prosperous cities in Nigeria (South West, South East and North West) using cluster approach. The results reflected on the link between consumer behavior and traditional clothing market, and provided guidelines for fashion store managers to improve their marketing strategies. Sentiment analysis through collection of responses (likes, shares, comments, re-tweets) helps the fashion store managers to analyze every aspect of consumers demand from the most loved colour to the most acceptable fits. Using these tools, the paper analyzed several designs to derive visual insight, producing a first-of-its-kind analysis of per-city fashion choices and spatiotemporal trends of modern civilization. Key word: Data Mining, Consumer behavior, Fashion style, Civilization and Cluster 1.1 Introduction Fashion industry is built around a process of producing a new look for people from time to time. The fashion business is an unending visual modification of the definition of society and its images. The motivation for using computer systems is clearly defined in the design, manufacturing phase to obtain the advantages of automated reliable designs that serve consumer best need. The computer is a natural medium for designing and mass production of perspective images; this makes it an effective tool for engineering, structural and stylish designers. In the recent years our economy had made a transition into the information age. Computers, data and information have become the basis for decision making in many industries, fashion inclusive. Companies have and are collecting very large amounts of information about their Customers, their products, their markets, their employees, their manufacturing processes, their distribution processes, and their marketing processes. This historical Information can be “mined” to develop predictive models to guide future decision. These ideas are being transitioned to Fashion industry in the form of new products. They have also become the basis for start-up companies developing entire new businesses. Through experience, an understanding has developed that data mining is a step in a larger knowledge discovery process. A systematic methodology has evolved to take raw data and transform it into information and to take information and transform it into knowledge to help solve real business problems. It is now understood that the larger process requires the careful use of different computer technologies in different stages. Raw operational data can be transformed into predictive models that support meeting major business objectives. Data mining plays a critical role in this overall process. These products correctly used can be successfully deployed into the business environment. There has been an evolution and development of very powerful and efficient database systems and associated access and query tools. Accord to Gul Kaner and Aykut Coskun (2017), Experts agreed that collaborative work between fashion and technology is essential to design fashionable, desirable and functional wearable technologies. They stated that they were willing to participate in such a collaboration. They shared their insights about stakeholders that should be actively involved in the collaboration, description of the collaborative product development process, the characteristics of collaboration environment and barriers for a successful collaboration. In the remainder of this section, we present these insights. 1.2 DATA MINING IN FASHION PERSPECTIVE Data Mining is defined as a process used to extract usable data from a larger set of any raw data. Data mining tools predict future trends and behaviors, allowing businesses to make proactive, knowledge-driven decisions, Madhuri V. Joseph (2013). The automated, prospective analyses offered by data mining move beyond the analyses of past events provided by retrospective tools typical of decision support systems. Data mining tools can answer business questions that traditionally were too time consuming to resolve. They scour databases for hidden patterns, finding predictive information that experts may miss because it lies outside their expectations. Over the years, fashion industries have had intuition at its disposal to predict customer demands which is now becoming relevant considering the fast-changing fashion trends and the tough competition in the market. More so, with more and more people getting brand conscious, it is becoming tougher for aspiring fashion designers to make a place on the mannequins. In a world where clothes become outdated with the release of a new movie or the latest fashion week, designers these days release photos of their exclusive collections on Social Media (Facebook, Twitter, Instagram, Pinterest) which helps them know the trends and people’s response much before the curtain-raiser. Sentiment analysis through collection of the responses (likes, shares, comments, re-tweets) helps the industry to analyze every aspect of consumers demand— from the most loved colour to the most acceptable fit. To support the business ideal, the ultimate data mining and decision support application might: I. II. Provide direct access to the data warehouse for on-line analytical processing Support rapid data extraction III. Provide for data quality verification and correction IV. Allow data cleansing, transformation, and preprocessing V. Provide multidimensional data visualization capabilities VI. VII. VIII. Provide a broad variety of data mining algorithms Generate useful and comprehensive reports Export actionable deployment strategies Furthermore such a tool would guide the analyst through each of the phases in the knowledge discovery cycle and enhance the effectiveness of the process. The company that successfully brings the concept of a full-service data-mining tool to market will advance the current state of the art to the next generation of data mining. 1.2 STATEMENT OF THE PROBLEM As sales, marketing and advertising become effective in the today modern innovative society, the present Nigeria market have little idea of how well billboards are working , we will be able to collect data on human gaze and hence the popularity of the brand in that area whereas other civilized society and cultures are using Big Data to improve its Social Media presence and Customer Engagement. Most importantly data mining also helps companies to optimize their supply chains as they can now decide what to produce more and should be stocked in inventory while what can be kept for made-toorder or Just-in-Time. Although, the use of Big Data may not completely redefine fashion industry as it is more of art, innovation and creativity than science and numbers, it definitely going to revolutionize the way industrialists and brands produce apparels and accessories will perceive instances. Intuition and innovation design are the basis of this industry but big data is what gives shape and direction to it leading to customers satisfaction. 1.3 RESEARCH OBJECTIVE The purpose of this study is to: I. To enumerate current uses and highlight the importance of data mining in fashion and Design field II. To find data mining techniques used in other fields that may also be applied in the Fashion sector. III. To identify issues and challenges in Data Mining as applied to the fashion and design practice. IV. Employed Data-mining method to analyze considerable amount of data collected from three cities as well as to understand the complexity of the diffusion process of multiple apparel products. V. To effect investment decisions 1.4 RESEARCH QUESTIONS In order to achieve the objective(s) of this study, the following questions were asked: I. II. III. Is there objectivity in promoting and propagating fashion as a profession? How do the media manage the business of fashion and its market? How can the customers be adequately managed in the fashion objectivism? 1.5 RESEARCH FRAMEWORK According to the results of the present study, style was not an important factor for the fashion leaders to purchase in South East, South West and North West. In terms of t-shirts and evening dresses/suits, 53% and 51% of fashion laggards are in the North West showing their strong preferences for fits and comfortable natives respectively. Additionally, 52% and 48 of the fashion leaders in South East and South West had shown a strong preference for fit and styles of T-shirts. Although this study is not exploratory due to society interconnectedness, we believe that data mining has great potential for investigating fashion diffusion of innovativeness and more replication of this type of research will be worthwhile and meaningful. 2.1 SENTIMENTS ANALYSIS IN THE PERSPECTIVE OF FASHION RETAILS Modern Fashion Retailers are innovating new ways to collect customer information and use it to provide a personalized shopping experience. While e-commerce itself is maturing day by day, customers still take it with grain of salt. There are aspects of it that can’t be changed due to its sheer nature. While customers have the opportunity to buy clothes from the convenience of home but there is no way for them to be 100% sure that what they are ordering will fit them. Fitting remains among the prime concerns of the customers when they shop online. And when not handled well, it's a source of additional cost for retailers as the orders with incorrect sizes from customers translate into support, and return requests. Social Media is democratizing luxury and fashion since customers are empowered with social media and all available data mining mechanism. In the traditional Fashion industry, what’s sold in the market comes from the taste-makers sitting at the top, telling the world what to wear. However, a big shift has started to happen as modern Fashion Retail companies have broken and changed the flow of fashion upside down with the help of cloud-sourcing. They have given customers a voice and enabled them to tell the brand what they want. Sentiment analysis is giving opportunities to customers to be part of business process. 2.2 CUSTOMER RELATIONSHIP MANAGEMENT (CRM) Fashion Industries and outlets maintain a huge volume of data about their customers and their request details. This information can be used to profile the customers and these profiles can be used for marketing and forecasting purposes. The emphasis of marketing application in Fashion Industries and outlets have moved from identifying new customers to measuring customer value and then taking steps to return the profitable customers. This shift has happened because it is expensive to acquire new customers than retaining the existing ones. A numerous Data Mining methods can be used to generate the customer life time value (the total net income a company can expect from a customer over time) for Fashion Industries and outlets customers. Different Data Mining techniques are used to model customer life time value for fashion customers, Freeman, E., & Melli, G. (2006). The key element of modeling the life time value for a Fashion customer is to estimate how long he/she will remain with their current trend. It will help the company to predict when a customer is likely leave and to take proactive steps to retain the customer. One of the serious issues that the fashion industries face is the customer churn. The process that a customer leaving a company is referred to as churn and churn analysis can be done through numerous systems and methods, Wei, C., Chiu, I (2002). Customer churn is a significant problem because of the loss of revenue and high cost of attracting new customers. Data Mining techniques are widely used for churn analysis. These rules were generated using SAS Enterprise Miner, a sophisticated Data Mining package that supports multiple Data Mining techniques. FIGURE 1 CRM Driven Data Mining techniques like classification, clustering and association rule are used for solving Fashion marketing business problems. Data Mining techniques are used for outlining four CRM dimensions namely customer identification, attraction, retention and development. Customer information and preferences can be used to determine customer behavior and identify the opportunities to support customer base expansion and customer churn reduction. Data Mining can be very useful in marketing programs development. For example if a service provider has a goal to increase the number of customers paying bills online or to increase revenue from advertisers then Data Mining techniques can be utilized. 3.1 METHODOLOGY This study gathered information through a questionnaire, Interviews and Observation. First, a detailed questionnaire based on the objectives of the study was administered to women and men. A random sample of 2000 Female and 1800 Male participated in the study. A total of 1900 and 1550 usable questionnaires were returned by respondents for the response rate of 85% and 86% respectively in each of the zones . 3.2 POPULATION AND SAMPLE This research work made use of population sample which comprised of Male and Female across the three selected zone South East, South West and North West. Proportionate stratified random sampling was used to select the required sample size of 2000 Female and 1900 Male retailers. The researcher used questionnaire, personal interview, Observation where appropriate to collect data and descriptive statistics were used in analyzing the data that relates to the research questions. The research instruments used for the Female subjects and the Male subjects were identical. 4.1 ANALYSIS OF DATA ON FASHION STYLE FIGURE 2 SOUTH EAST FIGURE 3 SOUTH WEST FIGURE 4 NORTH WEST reference Rosset, S., Neumann, E., Eick, U., & Vatnik (2003). Customer lifetime value models for decision support. Data Mining and Knowledge Discovery, 7(3), 321- 339. [6]. Freeman, E., & Melli, G. (2006). Championing of an LTV model at LTC. SIGKDD Explorations, 8(1), 27 -32. Wei, C., Chiu, I.: Turning Telecommunications call details to Churn Predictions: A Data Mining Approach. In Expert Systems with Applications 23, pp. 103-112. (2002). 4.2 RESULTS AND DISCUSSION This paper has presented an insights of the major fashion system regarding Collaborating with technology professionals. The results indicated that the experts are already motivated for such projects, and had a positive attitude towards collaboration with technology professionals. They indicated that strong collaborations are performed with individuals who have a personal and professional interest in working with technology thus, stakeholders’ willingness to participate is essential for a successful collaboration, Sanders, EBN (2000) The analysis of the adoption of Data Mining in the retail chain was based on a sequence of stages described in the system life cycle of the business Application. Each stage is described in greater detail here. [A.] PROBLEM DEFINITION Analysis of problems preceding the decision on Data Mining adoption in the retail chain was conducted by managers of the retail chain. Typically, managers will look for opportunities to improve the management of business processes in relation to the pressure on business processes efficiency (mostly due to increasing importance of e-commerce in comparison to the traditional fashion business) and the importance of the permanent need of innovation. The managers, mostly from the commercial department, faced several problems to be solved to achieve better efficiency: i. Share of information and analysis: The information system used in the company was not efficient, the particular systems were not connected (thus a problem in sharing information among departments) and analysis of the data from various aspects was lacking– production, sales, marketing, finance, logistics, and stock. ii. Decision-making: Managers’ decisions were sometimes delayed or changed due to lack of relevant data and information or other details. iii. Planning and pricing: The planning and pricing were not efficient due to imperfections in market change forecasts. iv. Stock optimization: There was the need to decrease the costs of stock management. v. The main objectives of the new system in the retail chain are to ensure that data are on time, accurate and appear in the required format, that the links of data from different functional areas are ensured and that shared information for better Cooperation of departments is available. This should result in improved forecasts of customer demands, planning, pricing, stock optimization, clear operatives, as well as strategic decision-making. [B.] FEASIBILITY STUDY The feasibility study examined the technical, economic and organizational feasibility of the project. It resulted in the decision that the outlet would adopt a new system, and, after cost analysis, decisions regarding the budget were made. The feasibility study was prepared according to the time plan, weekly, month, quarterly etc. The SAP Business Objects was chosen which had been expected to fit in with existing business procedures. [C.] REQUIREMENTS ENGINEERING All interviewees concurred that the stage of discovering and agreeing on exactly what the problems were and what the new system would do was the most demanding stage of the Business Intelligence adoption. The managers worked out a detailed list of key performance indicators (KPIs), including their definitions, calculation, input data and the influence of each KPI on company performance. [D.] DESIGN The technical solutions of Data Mining adoption (OLAP cubes and reports) and the data warehouse (DW) administration were realized in this stage. The subjects of business development were: the analysis, database design, ETL(Extract, Transform and Load) and reporting. After Implementation of business , administration became important: administration and maintenance of data warehouse, administration of ETL, database servers, reports and administration of complete business solution. The solution analysis focused mainly on the definition of data sources, inputs and outputs of DW and various functional areas, that are the basis for the functionality definitions included in a business object universe. The data flow from determined data sources to the implemented Business Object(BO) tool as illustrated in; 1. DATA LAYER Overview of the data sources: The following data sources were identified for the DW needs: a. internal ERP system, b. POS (detail data from individual stores), c. Shop Guard system – data on customer turnover, d. planning data gathered from the planning of the Business Intelligence project, and e. Manual CSV files. Some of the data were acquired through third party applications used in the company. All data from the source systems were entered into the DW through the defined input text files (CSV files). It was thus necessary to specify the path to the specimen file, destination directory and the frequency of data import. 2. INTEGRATION LAYER The most important parts of the integration layer were: i. DW in the integration layer, consisting of three databases: ii. DW (presentation layer) – historical and aggregated data in the form for reporting. The data are imported by an incremental approach in granularity and periodicity defined in the analysis by managers. iii. Stage (data transformation) – data are divided into dimension and fact tables. The data are changed and transformed for reporting. The tables on this layer contain original ID from source systems, as well as so-called “surrogate” keys. iv. Interface (source data) – between DW and source data. v. Ad-hoc ETL processes, which run three main tasks: importing the processing of the source data that can be imported in irregular time periods, plan actualization and maintenance of the data warehouse. 3. REPORTING LAYER The reporting layer was the most substantial part, as seen from the managerial perspective. The data areas had specific functions and also report values. The functional areas were: I. II. Sales, Shopping cart, III. Customer turnover and IV. Export. The facts were defined for functional areas and dimensions, including the hierarchy defined, for the purposes of reporting. [E.] IMPLEMENTATION Testing the system by the managers of the fashion retail chain will consist of six-member team from the retail chain (chief commercial officer, two senior category managers, supply chain manager, chief information officer and IT assistant) and developers from external company were responsible for the system implementation. A parallel conversion strategy was used, where the users had to operate both the old and the new systems. Deliverables from this stage of the life cycle included program listings, test plans and supporting documentation, details of the hardware on which the system would run, as well as manuals. [F.]MAINTENANCE Development of dashboards was the object of the maintenance of the system, something which could be dealt with earlier in the requirements of managers. This required additional maintenance time, approximately three months from requirement engineering until implementation, as well as incurring additional costs. 4.3 BENEFITS OF COMPUTER DATA MINING TO THE FASHION RETAIL CHAIN a. Commercially acquiring up-to-date and better quality information for decision-making regarding fashion trends b. Improvement of decision making (faster, better, based on better quality information as per arrival of trendy materials and designs in the warehouse or stores) c. supply chain manager: stock management – optimization, d. Marketing manager: improved ability to anticipate earlier changes on the market and e. E-commerce manager: better pricing. One category manager and chief commercial officer named information for decision-making as being the most important benefit. Four category managers agreed that improved decision-making was the most significant benefit of the new system. 4.4 SYSTEM MODEL FIG 5: The System Flow Diagram for Data Mining Operation 5.1 CONCLUSION The Data Mining Internet and related technologies has created an unprecedented opportunity for enterprises to collect massive amounts of data regarding customers and all aspects of their business operations. Yet the reality is that most organizations today are (i) “data rich” but “information and knowledge poor”, and (ii) not harnessing the full potential of their data, which is perhaps the second most important asset after human capital. Internet based applications such as social media, website usage tracking and online reviews as well as more traditional technology applications like RFID, Supply Chain Management (SCM), Enterprise Resource Planning (ERP) and Customer Relationship Management (CRM) provide access to vast amounts of data regarding customers, suppliers, competitors as well as a firm’s own activities and business processes. Being able to unlock the insights and knowledge trapped in such raw data constitutes a key lever for competitive advantage in hypercompetitive business environments. This paper is designed to showcase the virtually unlimited opportunities that exist today to leapfrog the competition by leveraging the data that organizations routinely collect every day, but which they hardly use strategically to make decisions at various points in the value chain. Students will be exposed to a wide gamut of issues related to data analytics and business intelligence, including the strategic aspects of big and better data as well as the details of analytical methods and data mining and visualization. REFERENCE Gul Kaner and Aykut Coskun (2017), Collaborative Design For Fashionable Wearables: A Fashion System Perspective.Nordes 2017: Design +Power,ISSN1604- 9705.Oslo, www.nordes.org Freeman E. and Melli, G. (2006), Championing of an LTV model at LTC. SIGKDD Explorations, 8(1), 27-32. Hakan Celik (2016) ,The Functionality of Online Shopping Site within the Customer Service Life Cycle: A Literature Review. Encyclopedia of E-Commerce Development, Implementation, and Management, DOI: 10.4018/978-1-4666-97874.ch05. Madhuri V. Joseph (2013), Data Mining and Business Intelligence Applications in Telecommunication Industry. International Journal of Engineering and Advanced Technology (IJEAT) ISSN: 2249 – 8958, Volume-2, Issue-3, February 2013 525. Sanders, EBN, (2000) Collaborative Design Proceedings of Co Designing, Springer, London Wei C. and Chiu I. (2002), Turning Telecommunications call details to Churn Predictions: A Data Mining Approach. In Expert Systems with Applications 23, pp. 103-112. (2002). Appendix A : Evening dress for ladies . Appendix B: Evening Dress for Men Appendix C: Fit Dress for Ladies Appendix D: Fit Dress for Men Appendix E: T-Shirts for Men Appendix F: T-Shirts for Ladies Appendix G: Native Yoruba Dress for Ladies Appendix H: Native Hausa Dress for Ladies Appendix I: Hijab Complemented Appendix J: Hausa Dress for Men Appendix K: Igbo Occasional Dress for Ladies