Data mining for retail website design and enhanced marketing

Data mining is considered as one of the most powerful technologies that participates greatly in helping companies in any industry to focus on the most important information in their data warehouses. Data mining explores and analyzes detailed companies transactions. It implies digging through a huge amount of data to discover previously unknown interesting patterns and relationships contained within the company data warehouses to allow decision makers to make knowledgebased decisions and predict future trends and behaviors. Industries such as banking, insurance, medicine, and retailing commonly use data mining to reduce costs, enhance functionality, and increase sales. Web mining is the process of using data mining techniques to mine for interesting patterns in the web. Those patterns are used to study user behavior and interests, facilitate support and services introduced to the website navigator, improve the structure of the website, and facilitate personalization and adaptive websites. In this dissertation, we developed a new approach that measures the effectiveness of data mining in helping retail websites designers to improve the structure of their websites during the design phase. This is achieved by giving them valuable information about the retail’s information system, its elements, and the relationships between different attributes of the information system. When considering this information in the design phase of the retail websites, they will have a positive effect in improving the website design structure. Furthermore, this approach reduces maintenance efforts needed in the future. We also studied the behavior of items with respect to time. This approach is beneficial in Market Basket Analysis for both physical and online shops to study customers buying habits and product buying behavior with respect to different time periods. We showed how association rule mining can be invested as a data mining task to support marketers to improve the process of decision making in a retail business. This is done through exploring current and previous product buying behavior and predicting and controlling future trends and behaviors. Based on our idea that interesting frequent itemsets are mainly covered by many recent transactions, a new method to mine for interesting frequent itemsets is also introduced. Finally, to solve the problem of the lack of temporal datasets to run or test different association rule mining algorithms, we introduced the TARtool. The TARtool is a temporal dataset generator and an association rule miner.