Quantitative analysis of the cities’ innovation capacity based on grey variable weight clustering

Abstract

Purpose – The purpose of this paper is to find the reason which the results of grey variable weight clustering method do not correspond with the reality. It proposes reconstructing the whitenization weight function, outlining why and how inconsistency is avoided. The study aims to improve the model of grey clustering method based on the whitenization weight function and list the steps of the new clustering model so that analysis and application of innovation capacity in a broader range is normally found. Design/methodology/approach – First the reason for the problem that the clustering results of grey variable weight clustering do not correspond with the reality is analyzed in two existing literature. And then a new whitenization weight function is reconstructed, two properties of the whitenization weight function are proved. The solution of the new grey variable weight clustering based on the whitenization weight function is built by following six steps. Findings – The paper provides a new whitenization weight function which satisfies the normative and non-triplecrossing. It suggests that successful clustering results of innovation capacity act on two levels: integrating the elements of innovation capacity indexes, and following steps of grey variable weight clustering. Originality/value – This paper improves the existing method of grey variable weight clustering and fulfills an identified need to study how cities’ innovation capacity can be clustered.

Portfolio management under capital market frictions: a grey clustering approach

“…Proceeding in a similar manner, by multiplying the difference by 0.66 and adding it to the lowest value, 3, we obtain 6.96, which represents the lower turning point of the f 1 2 function. As the boundaries of the three whitenization weight functions through which an object is classified in a cluster based on a particular metric have been arbitrarily chosen, and as the final classification of the objects depends on four metrics, we created a function for cluster B by allowing it to slightly overlap the functions for clusters A and C. This decision is in line with the literature (Xie 2014;Guan et al 2015;Dong et al 2018). As Guan et al (2015) state, in a clustering situation, one can encounter an overlap between two classes but never between three or more clusters.…”

“…As the boundaries of the three whitenization weight functions through which an object is classified in a cluster based on a particular metric have been arbitrarily chosen, and as the final classification of the objects depends on four metrics, we created a function for cluster B by allowing it to slightly overlap the functions for clusters A and C. This decision is in line with the literature (Xie 2014;Guan et al 2015;Dong et al 2018). As Guan et al (2015) state, in a clustering situation, one can encounter an overlap between two classes but never between three or more clusters. The area in which two classes overlap can be interpreted as a zone in which the value of the considered indicator might suggest that the object under investigation can belong to two classes at the same time.…”

Grey clustering and grey ranking of bank branches based on grey efficiency

“…Grey clustering and grey ranking are widely used approaches in various fields. Guan et al (2015) propose a quantitative analysis of the innovation capacity of cities based on grey variable weighting. They ameliorate the available grey variable weight clustering method and fulfill the identified needs to study how to innovate urban innovation capacity.…”

“…Grey clustering and grey ranking are widely used approaches in various fields. Guan et al. (2015) propose a quantitative analysis of the innovation capacity of cities based on grey variable weighting.…”