Last edited 11 Aug 2020

Clustering

ClusterExample.png

Contents

[edit] Introduction

Clustering is a method of pre-qualification for the procurement of construction projects. It applies the concepts of cluster analysis (which is a statistical technique used in many fields, including marketing, social science and computer science) as a classification method often used for data mining purposes.

[edit] What is cluster analysis?

When used in statistics, cluster analysis is used to divide data into groups (clusters) based on similar traits or characteristics. The goal is to identify traits that are meaningful, useful or both.

Cluster analysis refers not to one specific algorithm, but to the general task to be solved. The appropriate clustering algorithm and parameter settings depend on the data being gathered and the intended use of the results. There are roughly 100 different types - but one that is correctly selected can help to resolve an unwieldy or seemingly illogical decision making process.

Consider this example: there are a number of items in a room and each item has a set of characteristics (size, shape or colour). How will the items be grouped? Will it be based on size, shape or colour? This decision may be based on many factors, depending on how many items there are, their properties, priorities and so on.

[edit] Clustering analytics

Construction projects often attract a large number of potential suppliers with numerous characteristics.Under these circumstances, the prequalification process can sometimes require a significant commitment of time and resources.

The cluster analysis technique can be explored to identify common traits. These traits can then be used to group contractors based on their shared characteristics or abilities. Its purpose is to remove subjective rationale behind supplier selection and instead relies more on a pure data-based on evaluation.

Ideally, clustering organises suppliers by assigning values to key selection criteria. The end result is ideally a rational selection process based on evaluating each vendor, assigning a classification making a selection.

The analysis of data may include discerning trends and patterns, their interpretation and communication.

[edit] Related articles on Designing Buildings Wiki

Designing Buildings Anywhere

Get the Firefox add-on to access 20,000 definitions direct from any website

Find out more Accept cookies and
don't show me this again
"