Proceedings_____________________________________________________________________________________________________ of the Central European Conference on Information and Intelligent Systems 35 Delivery Optimization in Logistics Using Advanced Analytics and Interactive Visualization* Leo Mršić Algebra University College Zagreb, Croatia [email protected] Abstract. Paper is based on research and PoC with called a Eulerian Cycle. Finding such a cycle is more goal to unlock business value in delivery optimization, “feasible” and can be used efficiently to create business using advanced analytics and interactive visualization. value. This problem was first solved Mei-Ko Kuan, a There are several common approaches including Chinese mathematician, in 1962 [Kuan, 1962]. He was travelling salesman problem (TSP) that has various first one who consider this problem from the practical applications even in its purest formulation, such as perspective of a postman picking up mail at the post planning or logistics. Knowledge base powered by office, delivering it along a set of streets, and returning large data set is described through development and to the post office. Since he must follow route at least final, interactive, form. Using common principles, once, the problem is referred to as "Chinese" postman paper describe business value that can be extracted problem to investigate how to cover every street along using advanced analytics and interactive visualization, path and return to the post office under the least cost. how to structure steps as best practice to achieve In practice, there are a lot of applications of this success and what additional benefits can be expected technique such as road maintenance, waste collection, by supporting this approach. As part of research, bus scheduling etc. The objective of these problems various insights are extracted from PoC and presented which applied the model based on route optimization in forms suitable for general understanding and future is to find a route so that all of the edges of a given graph research. have been traversed at least once within minimum cost [Yilmaz at all, 2017]. Keywords. delivery optimization, knowledge base, advanced visualization, big data 2 Modern logistics 1 Introduction In the early days of computers, mathematicians hoped that someone would come up with a much better Delivery optimization is among most popular approach to large traveling salesman problems. research topics. Many practical routing problems Research community was looking for some algorithm involve finding paths or tours that traverse a set of that would allow computers to solve problem in a paths in a graph. The aim of solving such problems is reasonable amount of time. But while computer to find cycle with the least cost which covers all or a scientists have made progress with specific scenarios, subset of arcs in a graph with or without constraints volume of data was rapidly progressing: identifying the [Corberan and Laporte, 2014]. The Travelling shortest round-trip route for a 49-city map in the 1950s, salesman problem (TSP) and its variations are among a 2,392-city map in the 1980s and an 85,900-city map most popular central research problems in path routing. in 2006, putting pressure on algorithm that can The Travelling Salesman Problem (TSP) is about going efficiently solve every traveling salesman problem. to each point or node exactly once while returning to According to a landmark paper published in 1972 the original point or node (moving across in a cycle) [Karp, 1972], such a solution might not even be and also taking the shortest route among all possible possible. routes that fulfil certain criteria (if such a route exists). Modern technologies, especially big data tools and Finding such a cycle, perforce finding the possibly techniques, powered by mobile internet technologies unique optimal cycle with the shortest distance, is not made great success out of traditional services like ride- “NP-hard”. As popular variation, the Chinese Postman sharing systems such as UberPool [UberPool] and Problem (CPP) or Route Inspection Problem (RIP) is LyftLine [LyftLine]. Looking at model key business about visiting each route between nodes at least once value, it is obvious that they were based on allowing while returning to the original node and taking the multiple goals with similar itineraries and time shortest route among all possible routes that fulfil other schedules to share a delivering resource. By being able criteria (if such a route exists). A solution that takes to analyse and monitor various data (often real-time) each route exactly once is automatically optimal and such solutions and models can significantly increase ________________________ *This paper is published and available in Croatian language at: http://ceciis.foi.hr _____________________________________________________________________________________________________ 30th CECIIS, October 2-4, 2019, Varaždin, Croatia 36_____________________________________________________________________________________________________Proceedings of the Central European Conference on Information and Intelligent Systems delivery occupancy rate, alleviate traffic congestion, Table 2. Default time for different procedures and reduce the energy consumption of urban (selection) commuting [Ma et all, 2013] [Braverman et all, 2017] [Biswas et all, 2017][Zhang et all, 2014]. Default time Procedure Moving towards direction of modern logistics, for processing during analysis process two key modules in the ride- Collect post box 00:02:00 sharing system will be borrowed”: customer-vehicle Telegram delivery 00:04:00 matching and ride-sharing routing. The customer- vehicle matching module concerns with two issues: (a) Package delivery 00:10:00 how to group customers to create proper ridesharing Letter delivery 00:00:20 opportunities, and (b) how to match the formed groups to vehicles. We will modify this approach, confront Recent studies have been focusing on designing modern technologies with traditional approach and and optimizing the matching module [Braverman et all, identify business value unlocked during research 2017] [Biswas et all, 2017] [Zhang et all, 2014] so we process. will add postman delivery vehicle attributes as well. 2.1 Traditional KPI’s in logistic Table 3. Delivery vehicle attributes Route ID Vehicle Date/time Description Delivery management as part of supply chain is a RID-1 RN000 29.07. 14:11 Drive process that is often measured on its logistics costs. 29.07. 14:22 Stop That approach is a root cause of various issues but also RID-1 RN000 key point for business model efficiency. Various RID-1 RN000 29.07. 15:11 Drive computations do not measure the supply chain or its RID-2 TZ100 29.07. 10:11 Drive performance and can include factors outside of the supply chain. In addition, the way accounting treats Modern routing platforms aims to determine a path supply chain costs is dated. Fact is, delivery business is between nodes in the formed groups while different being transformed. The Amazon Effect has stimulated choices of paths can lead to different cost or profit to the beginning of a global supply chain revolution. It is the service providers. It is our aim to compare business moving beyond e-commerce/B2C and is crossing value between scenario where delivery travel demands industries and markets. The underlying expectations are given and delivery officer simply follow the path are pushing delivery business transformation which with minimum cost or maximum profit versus include improved performance and new metrics. possibility for same system to be updated with Supplementing the macro performance with segmented knowledge base data, preferably in real time. Main KPIs provides understanding and insight to what is difficulty of performing the matching and routing is happening on both centralize and "decentralized" that future demands are unknown when making views. It enables seeing underlying factors to the decisions. Existing works outline two approaches to "corporate" measure. addressing this difficulty. The first one is to assume As part of traditional metrics in postal delivery that all future travel demands are known at each include process steps standardization to quantify most matching decision timeframe, like offline approach valuable asses: time. [Biswas et all, 2017]. The second one is to assume zero knowledge of future demands [Zhang et all, 2014]. Table 1. Receipt of the shipment: steps Default time Step Type for processing 3 Delivery Optimization in Logistics Start/sort Package 00:00:10 Using Advanced Analytics and Logging in Package 00:00:30 Interactive Visualization Load Package 00:00:20 Unload Package 00:01:00 With the recent technical and scientific advancements in analytics lead to exponentially Package Document 00:01:00 growing sets of data. This generation of massive data Total time: receipt 00:03:00 along with the opportunities it provides for discovering new values and deriving new insights, and the various challenges it attempts to raise in terms of analysis and management, have created a new concept commonly referred to as Big Data [Borgi et all, 2017]. Logistics and transportation sectors are among the most ideally placed to take advantage of the analytical capabilities _____________________________________________________________________________________________________ 30th CECIIS, October 2-4, 2019, Varaždin, Croatia Proceedings_____________________________________________________________________________________________________