Page 132 - ASBIRES-2017_Preceedings

Asiri & Arudchelvam
the data into cubes. Data cubes can also be
used as an effective analysis tool. There are
many tools for data cubes which are
available as commercial products and they
are too expensive to allocate for the
research. Therefore, to develop a solution in
to a level where the client can be
compromised by proving the possibility of
making considerable change with the
solution, the solution development was
started based on the pure concepts of data
cubes where the actual development should
be done from the scratch. Following
possibilities are identified because that can
be developed from concepts
 Pre calculated data could be stored where
the storage for the row data will be
reduced as the calculation may take much
time
 Data can be fetched when needed without
calculations
 Increase the speed of searching
 No need of row data as the data are
available already calculated
 Unnecessary row data could be deleted
and this will free the storage
Regarding the development of the concept
the following parameters are identified.
Identified dynamic variables are:
 Time frequency interval - Per minute,
hourly, daily, weekly, monthly, and
yearly
 Action sources - Hotel Beds, TBX,
Bonotel, Pegasus, etc.
 Product Types - HTL, FLT, CAR, GEN,
TRN, etc.
Identified result parameters
 Average First Result Time – per action
source, per product
 Average Total Time - per action source,
per product
 Average Search Time - per action source,
per product
 Average Result Count - per action source,
per product
 Average Action time - per action source,
per product
122
For this development there were three
parameters identified. Those three
dimensions of a data cube are given below;
 Average rounded time
 Product type
 Action source
 After identifying these elements
It has been able to create one java project
to get all data using Elastic Search and
calculate values without using expensive
tools. Then a new index was created to store
these calculated cube data in the same java
project.
4 DATA COLLECTION AND
ANALYSIS
The project developed to implement data
cubes was tested with 10000 rows of data
and reported to be compressed to 3880 rows
including all the tested data in one cube
index. Therefore, at the testing, the storage
requirement was reduced with a
compression rate of 38.80% with the use of
data cubes successfully. Since the reduced
storage may lead to more efficiency in
searching, the performance can be improved
more than expected with the implementation
of NoSQL database. Using the generated
cube data, several graphs were created using
Microsoft Power BI tool. The generated
graphs prove that the generation of business
intelligence reports is effective with the
implemented solution. The Figure 1 explains
about the best action source which give us
best result count for a specific product type.
Figure 1: Graph indicating the action
sources and their average result counts for a
specific product type