A Data Mining Based Approach to Customer Behaviour in an Electronic Settings

Abstract

The understanding of customer incidents and behaviour is crucial to the success of any organization. Evidence from literature shows a prediction pattern of products to customer. These studies predicted product characteristics leaving out the customers characteristics. To address this gap, this study aims to design datamining system and implement it on an electronic commerce organization website. The customer information and history (clickstreams) from the electronic commerce website was used to predict the customers' behaviour. This will give meaningful and usable data patterns to organizations. Python programming language was used to design the datamining system, while PHP, HTML, and JavaScript were used for the e-commerce website. A brief description of the background of e-commerce and data mining, previous work of researchers who have worked on data mining in e-commerce settings, was reviewed and the relationship between their findings and this work was established. The data mining system utilizes consensus clustering technique and the clustering algorithm with a graphical-based approach. Furthermore, the interaction between the data mining system and the customer's dataset on an ecommerce website was defined. Quantitative evidence for determining the number and membership of possible customer behavioural clusters within the dataset was generated.

A Consumer Behavior Analysis Framework toward Improving Market Performance Indicators: Saudi’s Retail Sector as a Case Study

“…When dealing with a website, refs. [18,33] applied clustering algorithms to review the relationship between predicted consumer and product characteristics. Ref.…”

“…Ref. [18] extracted the appropriate patterns for a customer using web clickstream data, which the organization uses to produce a customized experience for the customer.…”

“…Some researchers directly observe and monitor customers' physical behavior [3][4][5], while others utilize specific data features to make predictions and analyze large-scale datasets to forecast consumer behavior [6][7][8][9]. Then, after approaching the consumer, a technical approach is implemented to predict consumer behavior, such as classification [3,5,8,10,11], association rule learning [9,[12][13][14][15][16][17], clustering [15,18], and other techniques like neural networks [4,6,7].…”