Clustering of Indian districts based on supply chain requirements. (2021)

• Record Type: Journal Article
• Authors: Baskar, A.
• Is Part Of: Materials today. Volume 46:Part 19(2021)
• Page Start: 9914
• Page End: 9919
• Publication Date: 2021
• Subjects: Facility Location -- Clustering -- Fermat-Weber Problem -- Weiszfeld's Algorithm; Materials science -- Congresses -- Periodicals; 620.1
• DOI: 10.1016/j.matpr.2021.02.292
• Languages: English
• ISSNs: 2214-7853

Abstract

Highlights: Proposes a model to develop a materials supply chain using geodetic coordinates. 661 districts (2011 census) of India are connected optimally. Weiszfeld algorithm is used to locate the cluster centres, iteratively. Both Manhattan distance and Euclidean distance are the distance metrics considered. Taking the population percentage share as the weights, clusters are computed in one case, which can be extended for other weights like GDP, literacy rate. Clusters can be computed by assigning any suitable weight. Up to five clusters are analysed and the optimum number of clusters is found. Abstract: Facility location problems refer to the selection and placement of a facility to best meet the intended requirements. The problem often consists of fixing manufacturing premises, process industry or office location that minimises the total weighted distances between the data points and the selected centre. The weights may be the constraints or preferences among the data points. The solution should comply with the stated or implied constraints and maximise the profit. Distance is one of the important constraints that have a direct impact on supply chain costs. This paper considers the 661 districts of India (2011 census) as the data points and finds the centres by clustering the districts into the predefined number of clusters. These centres and number of districts attached to each centre vary depends on the supply chain requirements. We can assume that the main customers are located at the district headquarters and products are to be transmitted from the centres with minimum time and cost to these points. Different algorithms are used for fixing a facility; based on the population, based on the distance and so on. It is assumed that the districts' headquarters represent the entire districts. Geodetic coordinates are collected for these 661 districts and Haversine formulae are used for converting them into earth-centric earth fixed (ECEF) × x, y and z coordinates. Using these coordinates, the popular Weiszfeld's algorithm is used in addition to four other implementations to solve and find the clusters and total distance among the districts in each cluster. All algorithms are coded in MATLAB 2012a and run in an i5 PC with 4 GB RAM.