Home | University of Tehran | Faculty of Engineering | ECE Department | CIPCE | Contact Us
DataBase Research Group - University of Tehran


Last Updated on: 21/07/07 
Data Mining
Title: Spatial Data Mining and its Application in bank business Intelligece

Work by: Ashkan Zarnani
Supervisor: Dr. Masoud Rahgozar
Advisor: Dr. Caro Lucas
Team Members: None
Problem Definition, goal and Importance: [persian]

Because of growing amount of data in enterprise systems and the growing need to get valuable information and knowledge from them, data mining has been introduced to satisfy this need. In fact data mining is a part of knowledge discovery which useful and hidden patterns are searched in it [3]. Growth in use of geographical information systems has provided access to huge data bases of this information. Data mining on data which has one or some spatial or geographical characteristics is called spatial data mining [1], [2], [4]. And the result is information which has spatial and geographical characteristics like, position, direction, distance, and etc.

Our goal is to perform a sophisticated data mining method on the spatial data in Mellat bank which will be mixed with information like, branch address, income, benefit, number of employees, and etc.
after preparing the data for data mining processes including Data processing and cleaning and building data warehouses, we must implement an algorithm for extracting the association rules and use it for finding the relations between different spatial values like, population, region, age, income, education, and situation of competitors, Besides banking index like income, benefit, and efficiency. Information extracted from this process can help the managers in their decisions.

Approach:

In this project we deal with quantitative fields and not Boolean nor categorical. So usual algorithms cannot be used and we must look for a method to solve the problem of ranges and number of them [6], [7]. Because of common specifications between geographical data, extracting relations from them should be done with considering some hierarchy of attributes which are created in a tree form are used for extracting rules. For example for topological attributes, tree is recommended. This hierarchy also can be used for geographical partitioning [5], [8].

Research Prerequisites:

Extracting association rules first was introduced by Rakesh Agrawal. He also had researches in quantitative fields. For the spatial databases and data mining in them, Jiawei Han has done extended researches for using common algorithms for data mining in those databases. He also recommended methods for using hierarchy in data mining.

Reports and Seminars:

Implementation:

Related Links:

Publications:


Copyright 2007 DBRG-UT. All rights reserved. Designed by Aresh Dadlani, Mohamad Hasan Ahmadi