Abstract:
One of the prime motives of information visualization is the analysis of time
oriented data. During last few years, a number of proposed solutions have
been published for the visualization of such data. There are other techniques
published for the multidimensional data as well as for large datasets. The
representation of multidimensional data is an ambitious task although there
have come a lot of methods for visualizing such data. Geographical data is
mostly represented using choropleth maps, but these maps are helpful if data
comprises of uni-variate or bi-variate data. For multivariate data these maps
are not helpful, the other representation techniques have to be considered e.g.
parallel coordinates, glyph based techniques, icon based techniques, texture
representation and motion charts etc. There can be number of techniques
for each dimensionality of data that represents data in an appropriate way.
In this thesis, a new method for the visualization of time series multivari-
ate geographical data is proposed. The multivariate data is based on some
scenarios in which data is divided into input and output streams. The visu-
alization is done using the size factor in which each attribute is represented
with this parameter; changes in value represent the change in size accord-
ingly. This technique is implemented using JavaScript library i.e. d3.js and
compare it with its established counterpart. An improved usability of 59% is
noticed after the experiment. A concise analysis of the proposed technique
is presented to open door for more effective visualizations.
