Abstract

Theme of master's work
"Development of information system
spatial data visualization"


Introduction
Topicality
Scientific and practical significance
Expected practical results
Clustering, the basic concepts and objectives
Classification of clustering
Measure the distance between objects
The clustering algorithm — k–means
Conclusions
References

Introduction

Geographic Information System is designed for the collection, storage, analysis and graphical visualization of spatial data and related information presented in the GIS objects. The term is also used in a narrower sense — GIS as a tool (software) that allows users to search, analyze, and edit digital maps, as well as additional information about the objects.

However, with increasing amount of data stored on the Internet, are having problems rendering a large volume of spatial data.

Nowadays, the most common geographic information system is Google Maps. But this system has its own problems with the mapping of large amounts of spatial data. The mapping GIS data can take a lot of time — even for high–speed Internet, such operations can be a serious challenge, not to mention the connection speed from the average user. One solution to this problem is clustering.

Topicality

At the moment there are a large number of clustering methods that use different measures and metrics. But despite this, the problem is acute, developing new algorithms and approaches. This problem is quite complicated, so is not completely solved, since for each task, you must select the appropriate algorithm and measure distances. The choice of metrics lies entirely on the researcher, because the clustering results may differ when using different measures.

Scientific and practical significance

Nowadays developments in clustering of spatial data by means of Google Maps almost underway. Since there are standard features of the Google Maps API. But in practice standard means clustering does not meet the requirements. Visualization of large amounts of spatial data takes a long time.

This problem will be solved by choosing a more appropriate clustering method, which will greatly improve the display of spatial data and make it easier to work with them.

Expected practical results

While working on the thesis work will examine the methods and clustering algorithms, to understand the scope of every of them and select the most suitable for this task. Result will be an information system that will solve the problem of visualizing large amounts of spatial data faster than the standard tools of Google Maps.

Clustering, the basic concepts and objectives

Clustering (or cluster analysis) — is the task of partitioning a set of objects into groups called clusters. Within each group must be "similar" objects, and objects of different groups should be different from each other. The main difference between the clustering of the classification is a list of groups that are not clearly defined and determined during the algorithm[4].

Cluster analysis performs the following tasks:

Regardless of subject matter, the use of cluster analysis involves the following steps:

Cluster analysis requires the following data:

After receiving and analyzing the results can adjust the chosen metric and clustering method to obtain optimal results.

Objectives clustering: