Developing an Interactive Dashboard to Improve Warehouse Performance

Developing an Interactive Dashboard to Improve Warehouse Performance Tracking and ABC Classification Final Presentation (Group 03) Chua Weilun | Russell Yap | Tay Jing Ying

Presentation Flow 1. Introduction 7. Dashboard Architecture 2. Problem Statement 8. Live Demo 3. Literature Review 4. Project Objectives 5. Data Provided 6. Methodology 2

Introduction • Home-grown supply chain solutions company • Operations all over Asia Pacific • Innovative logistics solutions to help improve the productivity of the labor-intensive logistics operations 3

Problem Statement • Unable to visualize or make sense of their current data. • Goods are put away in any available space in the warehouse • Fast-moving stock keeping units (SKUs) are not placed in optimal locations 4

Literature Review • Operational Dashboard will be a great COMMAND & CONTROL tool for warehouse managers (Eckerson, 2009) • Much literature on ABC analysis, but minimal studies talk about how to implement it. (Ashayeri, Heuts, Valkenburg, Veraart & Wilhelm, 2002; Montulet, Langevin & Riopel, 1998) 5

Literature Review • Market Basket Analysis (MBA) aims to find association rules from transaction data. • Can be used for scenarios other than sales. (Sokołowska, 2014). • Use Graph Network Diagram to allow users to conduct their own exploration of the MBA analysis 6

Project Objectives • • Develop an Operations Dashboard to visualize: § Movement volume of inbound and outbound goods § Seasonality / Ranking of products § Location of each category (A, B or C) of products Conduct Associative Rule learning to find affinities within products to better organize inventory • Recommend locations within the warehouse to place each product 7

Data Provided DATASET • Company which sells cables and connectors • Provided with data from 2015 to 2017 • Two datasets

Data Provided Warehouse Layout – Rows and Lanes Example Loc. Name: 12 D 32 M

Data Provided Example Loc. Name: 12 D 32 M Warehouse Layout – Shelf Levels

Methodology • ABC Classification • Market Basket Analysis • Warehouse Recommendation Model 11

ABC Classification • An inventory classification technique used by warehouse managers • Classifies inventory into 3 categories: A, B and C • The most general way of categorization is: § ‘A’ items - 20% of the items accounts for 80% of the annual consumption value of the items § ‘B’ items - 30% of the items accounts for 15% of the annual consumption value of the items § ‘C’ items - 50% of the items accounts for 5% of the annual consumption value of the items 12

ABC Classification Aggregate data from order level into product level with PRODUCT_CODE as the key and the sum of NO_Of_CARTON as its value. e. g. PRODUCT_ID | NO_OF_CARTON 1042 | 2000 Sort the list using the total number of cartons from HIGHEST to LOWEST Create 1 more column called cumulative frequency. This is done by using the Relative Frequency of the current row and adding it to the previous rows cumulative frequency 13

ABC Classification 14

ABC Location Classification Methodology • Idea: MINIMIZE the overall time needed to pick products • Allocate products in the warehouse based on its ABC Classifications • Assign locations in the warehouse with a classification based on its distance from the packing area. 15

ABC Location Classification 16

ABC Location Classification Before Removing Duplicates After Removing Duplicates 17

ABC Location Classification 18

ABC Visualizations Objectives • Allow users to visualize the actual status of the warehouse • View how the optimal situation should look like 19

ABC Visualizations Visualization – Warehouse Heat Map 20

Market Basket Analysis • A way of using transaction data to find out the RELATIONSHIPS between the items that people buy. • Usually used in retail sector to come up with bundles or to help plan out the layout of product placements in a store. • Main purpose is to derive ASSOCIATION RULES • {Milk} {Beer} 21

Market Basket Analysis Apriori Algorithm ▫ Purpose is to ELIMINATE transactions which do not satisfy the MINIMUM occurrence threshold ▫ Use remaining item to form valid item sets which will be become ASSOCIATION RULES 22

Market Basket Analysis 3 Key Metrics 1. Support 2. Confidence 3. Lift 23

Market Basket Analysis Support • Proportion of orders in the whole dataset that contains the item set • Support {LHS} {RHS} or Support {LHS, RHS} = (number of transactions containing LHS and RHS) / (number of all transactions) 24

Market Basket Analysis Confidence • The probability that a transaction that contains the LHS of the rule will also contain the RHS • Confidence {LHS RHS} = (number of transactions containing LHS and RHS) / (number of transactions containing LHS) 25

Market Basket Analysis Confidence • The probability that a transaction that contains the LHS of the rule will also contain the RHS 26

Market Basket Analysis Lift • Indicates the existence of a relationship between two items • Lift {LHS RHS} = Lift {RHS -> LHS} = (number of transactions containing LHS and RHS) / (number of transactions containing LHS) * number of transactions containing RHS 27

Market Basket Analysis Lift ACTUAL Probability of Apple and Egg occurring together EXPECTED Probability of Apple and Egg occurring together by PURE CHANCE 28

Market Basket Analysis Lift • Lift = 1 implies NO relationship between A and B • Lift > 1 implies that there exists a POSITIVE relationship between A and B • Lift < 1 implies that there exists a NEGATIVE relationship between A and B 29

Market Basket Analysis Methodology ▫ Prepare data into suitable format 30

Market Basket Analysis ! S TE A C I PL O M RE U D VE 31

Market Basket Analysis Methodology ▫ Remove Duplicates 32

Market Basket Analysis Visualization – Graph Network Diagram • Nodes = Products • Edges = Links between products • Experiment using Gephi 33

Market Basket Analysis Visualization 34

Market Basket Analysis Visualization 35

Warehouse Recommendation Model Methodology • Combine results of ABC Classification with the output of MBA analysis to achieve 2 objectives 1. Allocate incoming products into their OPTIMAL LOCATIONS 2. Placing products as close to its AFFILIATES as possible 36

Warehouse Recommendation Model Implementation – 2 checks ▫ ▫ Find valid locations (ABC Classification) ▪ Is it in the same category? ▪ Is there available capacity? Is there a an affiliate within the valid locations? (MBA) 37

Warehouse Recommendation Model Implementation – MBA Check ▫ Products which are placed further away from a strong affiliate will be penalized 38

Warehouse Recommendation Model 39

Dashboard Architecture 40

Live Demo! 41