logo

Study on Market Basket Analysis Technique for Inventory Management

   

Added on  2020-09-30

147 Pages35749 Words213 Views
Project Title: Study on Market Basket Analysis Technique Page 1 of 147for Inventory Management _____________________________________________________________________________________________ Prepared by: LEE ZHEN XI FYP4201/4203 Project I January 2019 SessionFINAL YEAR PROJECT REPORT PROJECT TITLE: Study on Market Basket Analysis Technique for Inventory Management "A dissertation submitted in partial fulfillment of the requirement of an Honours Degree in Computer Science / Information Technology at INTI International University under the management and supervision of Faculty of Information Technology I declare that this project is my own work, it has not been copied in part or in whole from any source except where duly acknowledge. As such, all uses of previously published works (from books, journals, internet, etc.) have been properly acknowledged within the report to an item in the references or bibliographies. I hereby submit my dissertation, dated 24/04/19 for review and assessment. by Student Name : LEE ZHEN XI Student ID : I15008178 IC No/ Passport No : 971207-05-5343 Project ID : FIT-INTI-IU-BCSI-JUNE-2019-0025

Project Title: Study on Market Basket Analysis Technique Page 2 of 147for Inventory Management_____________________________________________________________________________________________ Prepared by: LEE ZHEN XI FYP4201/4203 Project II January 2019 SessionACKNOWLEDGEMENTS First the author would like to thank to his family, especially his parents who giving, caring and lovely when the author is stressful in doing the stage of developing the documentation and algorithm process. Secondly, the author would like to thank to his project supervisor, Ms. Hema Latha for her guidance along with this project. Ms. Hema had given me a project title and guidance on my the project. Besides, she will meet me for the consultation on every progress on the proposed research. She did help in correcting the mistakes and grammar during the documentation process in chapter one. Lastly, the author would like to thank to his classmates and everyone who been helpful during the development progress. Their willingness to answer the interview questions and giving suggestion on my research. The authors appreciate at the time they spend in teaching me with the new knowledge which needed for the project.

Project Title: Study on Market Basket Analysis Technique Page 3 of 147for Inventory Management _____________________________________________________________________________________________ Prepared by: LEE ZHEN XI FYP4201/4203 Project I January 2019 SessionAbout the Author Lee Zhen Xi is the author of this project, he had studied on Diploma in Information Technology in the major of Computer Science. Currently, he is studying Bachelor of Computer Science with major of Business Analytics. He has worked as an intern in the IT department at Asia Tradex. He has a knowledge of data mining and predictive analysis and business intelligence. He has attended the classes and able to complete his assignment with a time given and he has achieved the objective of each assignment.

Project Title: Study on Market Basket Analysis Technique Page 4 of 147for Inventory Management_____________________________________________________________________________________________ Prepared by: LEE ZHEN XI FYP4201/4203 Project II January 2019 SessionResearch Proposal ________________________________________________________________ Title:Study on Market Basket Analysis Technique for Inventory Management__________________________________________________________________ Abstract: The aim of the project is to study on data mining techniques and to find the best market basket analysis techniques. Market basket analysis is useful as it allows retailers to find purchaser habits and it encourage the retailers to make the right decision on the stock arrangement and inventory management. Market basket analysis able to discover the connection between groups of the items, products and classifications. Most of the company who is under the categories of retails shop and e-commerce are having a poor sale. The store product arrangement is not optimize based on customer need therefore market basket analysis will be the good technique to identify how arrangement of the product. Static data that is available from the retail is not able to capture buyer behavior therefore proposed algorithms not only mine static data but also provides a new way to consider changes happening in data. The Data Mining technique and similar application who works on market basket analysis will help to get the live data and compare with their business objectives. It will be catering the needs of the both consumers and enterprises and reclean the data to useful data so that it can help in unlock the new revenue streams __________________________________________________________________

Project Title: Study on Market Basket Analysis Technique Page 5 of 147for Inventory Management _____________________________________________________________________________________________ Prepared by: LEE ZHEN XI FYP4201/4203 Project I January 2019 SessionResearch Description: Introduction The main purpose of market basket (association) analysis is to identify co-occurrence relationships and discover logical rules that describe associations between variables in a dataset by identifying positions where they co-occur. Market basket analysis also known as association-rule mining is a useful method of discovering customer purchasing patterns by extracting associations or co-occurrences from stores’ transactional databases (Chen et al. 2005). The technique originates from the analysis of transactional data generated by retail chains. The name of its most common application, Example market basket analysis refers to the customer’s shopping basket. The method enables shop owners to make sales management decisions, plan production, promotion and arrange products in the most optimal places. It also helps the marketing analyst to understand the behavior of customers, e.g. which products are being bought together (Kaur, Kang 2016). Over the years, market basket analysis has started to play an increasingly important role in the analysis of financial and insurance transactions (Roodpishi, Nashtaei 2015). Problem Statement Each transaction occurs in a single time period, where W= {W1, W2, W3.....WT} is the set of all time periods under consideration. Time periods have the following assumption. Collectivity is for one transaction per store, per user can occur in a time period. The Collectivity assumption takes into consideration the practical aspects of shopping and builds on the Quantity assumption outlined for items. The generally accepted length of a shopping period is one week as it is in line with how most people plan their household activity. Hence all items purchased within the week may be considered to be the market basket of the household. (Kantar, 2017)Given this, it then becomes important to identify all items purchased in a given store in that week, hence the day or time of purchase is immaterial. Consequently, all items purchased from the same store in the same week is aggregated into a single transaction. Having too short a time period is not practical for a sound analytical exercise as it will result in delinking of obvious shopping patterns. For example, if the period was one day then the model will suggest that items bought by the same user on Saturday in store “S1” are not related to items bought on Sunday in store “S2”, for example. In practice this is not the case and consumers will generally spread their grocery shopping over the weekend or several days based on logistics. (Manthan January 30, 2017.).

Project Title: Study on Market Basket Analysis Technique Page 6 of 147for Inventory Management_____________________________________________________________________________________________ Prepared by: LEE ZHEN XI FYP4201/4203 Project II January 2019 SessionThe assumption on Quantity is a necessary simplification and whilst it does impact frequently purchased item sets in that bulking buying of a single product in one time period could make the overall frequent itemset less‐frequent over several time periods, this behavior is becoming more an exception that the norm given the changing shopping patterns of today’s consumers.(Ellis‐Chadwick F, Doherty NF, Anastasakis L 2017) A transaction is a set of all items purchased and is recorded in a database. There are three dimensions to a transaction namely: store, user and period. Hence a transaction is fully detailed as Ts,,u,,t=I2I5.....Iietc and may be read as “A unique transaction “T” occurring in store, “S”, for user, “U”, in time period, “T”, contains items I2,I5... etc.” Note that the database is binary with the physical quantity of each item purchased being ignored. For simplicity the time period is usually a week and will be ignored, unless it is relevant. A universal transaction database, D0, represented by S=0, is obtained by combining the transactions for all users in a given time period across all stores. Hence “T0,u,t=T1,u,tU T2,u,tU...U Ts,u,t” etc. is a combined transaction for user, “u”, and exists in “D0”. This may be seen as the user’s view of their transacting in a given period across all stores. In practice this universal database does not exist for all users, but large subsets exists through the consumer scanner panel programs run by third party analysts like(Kantar January 30, 2017.) which are then used to make inferences of the overall market. It should be noted that T0,u,tis a binary transaction, consequently the same item “Ii” purchased from multiple stores in the same period by the same user will only be counted once as the database denotes “presence” as opposed to quantity. Research Questions 1. What is the customer buying behavior? 2. How to identify the business target segments? 3. How to identify the strategic floor plans of a shop? Research Objective •To identify the consumer buying behavior for decision processes and the transaction record of one customer •To identify the business target segments according to previous sales volume or potential sales volume. •To identify the strategic floor plans of a shop who meet the needs of buyer and able to improve the sales of another product

Project Title: Study on Market Basket Analysis Technique Page 7 of 147for Inventory Management _____________________________________________________________________________________________ Prepared by: LEE ZHEN XI FYP4201/4203 Project I January 2019 SessionResearch Purpose In the journal (Ellis‐Chadwick F, Doherty NF, Anastasakis L 2017), it questions on “What does beer have to do with diapers?” It was a common question until one of the largest US retailers conducted a study and found that diaper sales were associated with beer.( Ellis‐Chadwick F, Doherty NF, Anastasakis L 2017).In general, buyers were men who went out at night to buy diapers and ended up buying some beer. The products were placed side by side and the result was increased sales of the two products. (Ellis‐Chadwick F, Doherty NF, Anastasakis L 2017) It is unknown whether this study actually occurred or whether it is only a "legend" about data mining study. The competitiveness of companies is becoming increasingly dependent on the quality of their decision-making. Therefore, it is no wonder that companies often try to learn from past transactions and decisions in order to improve the quality of decisions made in the present or future. To support this process, large amounts of data are collected and stored during business operations. Later, this data is analyzed for relevant information. This process is called data mining or knowledge discovery in databases. Data mining is relevant to many different types of companies. As examples, retail stores obtain customer profiles and their purchasing patterns and supermarkets analyze their sales and the effect of advertising on sales. This "targeted marketing" is becoming increasingly important. Thus, this case study tends to propose a specific group of products purchased by customers of a financial institution and predispose a tying or otherwise known as cross-selling, ie, point where the product this customer is likely to purchase, by consumption profile. (Ellis‐Chadwick F, Doherty NF, Anastasakis L 2017). Association rule mining and Business intelligence concepts are used to find interesting association or correlation relationships among a large amount of data items. With huge amounts of data constantly being collected and accumulated, many organizations are showing interest in mining inventories and handling them effectively from this large collection of business transaction records, as it can assist in many business decisions making processes, such as catalog design, cross marketing and others. And the smart inventory management can be used to improve the business level and profit. Finding and calculating the inventory aging and anomaly entry from large databases is extremely convoluted. These databases contain numerous irrelevant and redundant data records which are not essential to extract the desired results. In addition, these irrelevant data considerably affect the quality of the inventory miner and hence there is a requirement to preprocess these records. This will be performed using effective feature selection algorithms. Mining inventories with its aging and demand forecasting in large databases play a vital role in the field of market analysis. The rules which are generated using association rule and inventory mining tools can help to the user make the right decision on the market data. This paper addresses the above problem along with some research issues such as the issues associated with the discovery of frequent, closed and weighted item sets from transactional data sets for effective demand forecasting, which also finds the anomaly in inventory streaming. The problem of

Project Title: Study on Market Basket Analysis Technique Page 8 of 147for Inventory Management_____________________________________________________________________________________________ Prepared by: LEE ZHEN XI FYP4201/4203 Project II January 2019 Sessionmining high demand products by considering effective data mining is the popular research area. Every algorithm proposed in literature results in slow computation due to its uncertain huge datasets. (Venugopal, Jisha. P. 193 - 196. 2017.) As outlined in (Platzer and Reutterer (2016) , one of the most challenging areas remains the prediction of customer purchases in the non-contractual settings: The current status of the customer is not directly observable at a time and the available historical record is censored while customer data tends to vary substantially. During the last years, large improvements in the information technology domain have resulted in the increased availability of customer transaction data (Fader & Hardie, 2009 ). Initial analyses of these transaction databases are usually descriptive in form of basic summary statistics such as the average number of orders or the average order size, and information on the distribution of behaviors across the customer portfolio. Further processing of the customer base may use multivariate statistical methods and data mining tools to identify characteristics of, for instance, heavy buyers, or to determine which groups of products tend to be purchased together (i.e., performing a market-basket analysis).The purpose of this project is to identify the consumer buying behavior which include their attitude towards consumerism, beliefs, purchasing patterns and behaviors by figure out this, we will be able to convince the customer on buying the product or service. From the data of the retail shop we will able to identify who isn’t a prospect for the product and to look at what the customers have in common to identify which customer cause the minimal service problem and maximum profits. For that purpose, we first compute a large number of customer features, that characterizes the customer at a given month. for predicting whether the customer makes a purchase in the upcoming month. The application of data mining techniques for predictive purposes on the customer base is often analyzed in the customer relationship management and expert systems domain, and customer churn prediction is the most popular objective in this fields.

End of preview

Want to access all the pages? Upload your documents or become a member.

Related Documents
Dissertation | Nan Tien University - Case Study
|31
|4859
|29

Effectiveness of the Current Health and Safety Policies
|54
|13333
|186

Skilled Birth Attendance (SBA) | Research Report
|28
|8675
|22

CSR In The Delta Niger Region Thesis 2022
|53
|13774
|22

Impact of Social Media on Consumer Buying Behavior in Woolworths Limited, Australia
|61
|12237
|205

Employee Development Programs on Job Satisfaction Assignment
|56
|15661
|99