Reporting and Decision Support Using Data

Warehouse for E-Commerce top-up Cell-phone

Credit Transaction

1st Vito Vincentdo, 2nd Andhika Rizky Pratama, 3rd Abba Suganda Girsang, 4th Rizaldi Suwandi, 5th Yosefina Pricilia Andrean
Computer Science Department, BINUS Graduate Program-Master of Computer Science
Bina Nusantara University
Jakarta, Indonesia 11480
[email protected], [email protected], [email protected], [email protected], [email protected]

Abstract—As e-commerce that covered all Indonesia, needs II. RELATED WORK
a fast reporting system to analyze the business to make the best
decision. This paper provides methods to fasten the reporting Business Intelligence (BI) involves several distinct areas
process by making a data warehouse using Kimball and ross and technologies that converge in the common goal of having
method. Data warehouse consists of 9 step methodology access to data in order to help businesses in other words [5],
designed by Kimball and Ross. This paper devoted for business
processes in e-commerce especially for online cellphone balance BI can be defined as the process of turning data into
top-up transactions. Furthermore, data can be displayed in a information, and then into knowledge [6]. According to Alpar
form such as a dashboard, and also in a simple, informative and P., & Schulz, M. Business Intelligence (BI) systems had
effective report.
undergone two distinctly different fundamental changes. On
Keywords— Data Warehouse, Kimball, E-commerce, Online the other hand, social media systems, machine sensors,
Transaction devices like smartphones, and other sources generates new
data which often differ from traditional operational data
I. INTRODUCTION regarding their structure, rate of growth, and volume. The
XYZ is an e-commerce company that sells a variation of scope of BI has been extended from strategic questions to
products start from everyday grocery, industrial, medical, operational tasks so it’s very important now for every
electronics digital products and many more. One of the organization to use Business Intelligence because it can
products that is sold by XYZ is cell-phone credit. From each improve efficiency and give a good result of increased
transaction that occurred, each of them will be recorded in a productivity, it can also cut the organization expenses.
database. These data doesn’t have any meaning yet and Business Intelligence uses a variety of tools or applications to
doesn’t have any other role besides for reporting. Report that retrieve data from the warehouse and to present it to users.
generated from each transaction usually will generated in a Process from BI can be used for making business decisions
monthly period. Report still in excel form and takes a moment in an effective and efficient manner [7]. For example, user
to generate a report from some excel file. can use BI in the supply chain, sales, finance, and marketing
[8]. The key characteristic of this, is not unique to Business
The occurring problem can be solved by using a data Intelligence but certainly more prevalent in Business
warehouse system. As defined by John L., data warehousing Intelligence culture than in traditional BI, the reports are often
technology is a set of methods, techniques, and tools that are interactive with their users, which may include proactive,
used to create a vehicle that delivers data to end-user on an responsive, and contextual. More employees have a need to
integrated platform [1]. The data warehouse can also know as apply BI [9].
a database that stores a collection of data that exists and
present data. Each data comes from different sources and Data warehouse is a relational database system that is
formats. Data warehouse integrated some source of data to be used to store and analyze data, also report the results to users.
a single formatted consistent database. The challenge in the The data mart is based on a data warehouse and facilitates the
making of the data warehouse is to withdraw data and access and analysis of the data warehouse. A data warehouse
combine those data into 1 structure through an ETL (Extract, is mainly responsible to store data that is sourced from the
Transform, Load) process [2]. The methodology used for data operational systems [10]. Data Warehouse Methodology,
warehouse development is Kimball lifecycle methodology Methodology itself means, a system that had been used in a
which published in the 1990s. This methodology was used particular study/activity, in this case, is software engineering,
and proven by many researchers in various fields [3]. This data warehouse to be specific, in software development
design had a well-known paradigm in the area of data methodologies such as waterfall, spiral, Rational Unified
warehouse and database in general, always related to OLAP Process (RUP), Agile, and etc. [11]. One of the methods that
tools [4]. was used is the Kimball method, and it’s using the bottom-up
approach which is much better for data reporting purposes.
The data warehouse has a big role in every organization to
analyze what’s trending in their data within a certain period
of time to stay competitive. From recent studies, key factors
The 7th International Conference on Cyber and IT Service Management (CITSM 2019)

to a successful data warehouse is, improved availability of data warehouse. Dimensional model of any business process
data to meet business needs; improved capability of satisfying has two distinct aspects, measurement, and its context. In data
analysis and reporting needs; improved data quality for better warehousing, this term is known as fact and dimension [15].
decision-making; and improved timeliness of information for
A. The Dimension Table
business users use [12].
The next step after completing star schema is data
III. PROPOSED METHOD processing. First, decide which one to process from the
The Kimball methodology involves a singular focus on simplest table such as product, source, status, and user
individual business processes as the basis for designing dimension. All of this table didn’t need any complex
analytical reporting database structures. This is its strength, transformation. The process was, extract from its source,
but also a weakness, as it does not look at how these processes select a value that correspondence to the target field, and load
interrelate. The interrelation of processes and associated to the target table. Fig. 2 is one example of the process that
information is important in areas such as health research that occurs.
need to examine cause and effect [13].
Based on bottom-up data warehouse by the Kimball
approach, there are 4 steps of development. The first is to
select the business process, this requires a full understanding
of the process that happens inside this company. To
understand this, analyzing every possible transaction that Fig. 2. Source Dimension ETL
could happen in any scenario is needed. Second is to declare
the grain of a fact table. Grain is what do you want to describe For the date and time dimension, the ETL was slightly
from a single row of the fact table and determines the level of different. Calculator transformation from Pentaho was used
detail. The third is to identify the dimensions needed to make to calculate the transformation. This transformation
the data warehouse system running. Dimension typically can simplifies complex query process into a single function. For
easily identify end as they represent the “who, what, when, ETL process in Fig. 3, we used calculator to extract a date,
why, where and how” associated with the events. The last month, year, and quarter from a date and load into the target
step is to identify the fact. The fact can be described by a table table. The time dimension is not so different from date
that store a specific transaction, in this case, is online top-up dimension, it is used for the same transformation to extract
cell-phone credit. The end result from this step is to produce the hour and minutes. The hour and minutes can be extracted
a star schema as shown in Fig. 1. when the date source format field includes hour and minute.

Fig. 3. Date Dimension ETL

The last dimension is payment dimension. This dimension

is a little more difficult to make because it needs integration
between 2 different tables into 1 product dimension. The
process in Fig. 4 is joined payment method and payment
detail table by lookup the detail from the fact table.

Fig. 1. Star Schema Model for Online Top-up Cell-phone Credit

Transforming data conceptual model into a model based

on a large structure table approach has been used, which Fig. 4. Payment Dimension ETL
keeps the fact and the dimension joined. This approach can
B. The Fact Table
be called a denormalized logical approach (DLA) [14].
From completing the dimension table ETL, the next
IV. ANALYSIS RESULTS process is to make the Fact ETL. In this case, only 1 ETL
As seen from star schema in Fig. 1, there are 8 tables. needed to make the fact table. From Fig. 5., can be seen that
Each of these tables is generated from an ETL (Extract, the old transaction fact table was processed and transform
Transform, Load) that processed data from an excel file, into into data warehouse fact. The difference between old fact and
an integrated database table. There are 8 ETL process, 1 is the the data warehouse fact are the contents that kept inside. For
fact process and the remainder is the master table process. fact in a data warehouse, A key from each dimension are
Pentaho Data Integration Service has been used to make this stored. This is to minimize the redundancy and also minimize
 The 7th International Conference on Cyber and IT Service Management (CITSM 2019)

the impact if there is a change. The data in dimension and

data integration can easily be changed into fact.

Fig. 5. Payment Dimension ETL

C. Generating Report
By finishing all the ETL process now data warehouse is Fig. 7. Cell-Phone Credit Sales Dashboard
ready. The last step is designing the reporting form that
needed to process. For each design, report design only Fig. 7. Shows the dashboard of company XYZ Cell-Phone
depends on the requirement analysis. For example, designing Credit Sales. Many things can be identified from this
a report that can show sales rates based on data warehouse. dashboard such as the most selling product, the most loyal
customer, the best sources, and total sales of Cell-Phone
Credit. The First one is Sources Bar-Chart, from this chart the
best sources can be identified, this might be useful for future
development. The second one is a pie chart that shows the top
5 customers so the company knows who is the most loyal
customer. The Third one is the selected product. This
selection will affect all of the output shown by the other item.
The Fourth one is Product Bar-Chart, from this chart the Cell-
Phone Credit Sales per-month can be seen. And the last one
is a KPI (Key Performance Indicator) that will show the total
sales of Cell-Phone Credit Dashboard that might be useful for
the company to evaluate their success at reaching targets.
V. CONCLUSION
With the growth of size and variety of data, it is important
to make the right choice on how to manage the data. It is
important to build a structure that can handle and cover all the
Fig. 6. Report and Chart for Specific Total Product Sales
requirements. Based on the reporting and decision support, a
data warehouse can handle and also make it easier and faster
Fig. 6. shows the sales report of a specific product and to generate reports and decision support. Data sales from all
summarizes it within a month. This report was generated providers and nominals have been processed. For future
from the data warehouse database and design it into a specific works, there will be a prediction for the top three products
and simple design. From this report, hopefully, the company and nominals that will hold in sales for 1 quarter. The results
analyst can compare the report and make a decision to from these data can be used by the business team to manage
improve the company operationally. or prepare a strategy and can be a spear to get revenue.
D. Dashboard REFERENCES
The dashboard is an information management tool that [1] S. T. Su and A. Needamangala, “Harvesting Information from a
visually tracks, analyzes and displays metrics, key data points Library Data Warehouse,” ITAL, vol. 19, no. 1, pp. 17–28, 2017.
[2] A. S. Girsang et al., “Decision Support System Using Data
and key data performance that used for monitoring the health Warehouse for Hotel Reservation System,” SIET, pp. 369–373,
of a business. Users will only see all the data in the form of 2017.
tables, line charts, bar charts, and gauges. [3] S. N. Murphy et al., “Combining clinical and genomics queries
using i2b2 - Three methods,” PLoS One, vol. 12, no. 4, pp. 1–16,
2017.
[4] A. Abello et al., “Using Semantic Web Technologies for
Exploratory OLAP : A Survey,” IEEE Trans. Knowl. Data Eng.,
vol. 27, no. 2, pp. 571–588, 2015.
[5] S. Moro, P. Cortez, and P. Rita, “Expert Systems with Applications
Business intelligence in banking : A literature analysis from 2002
to 2013 using text mining and latent Dirichlet allocation,” Expert
Syst. Appl., vol. 42, no. 3, pp. 1314–1324, 2015.
[6] M. Golfarelli, S. Rizzi, and V. Castenaso, “Beyond Data
Warehousing : What ’ s Next in Business Intelligence ?,” DOLAP,
pp. 1–6, 2004.
[7] M. S. Sangari and J. Razmi, “Business Intelligence Competence,
Agile Capabilities, and Agile Perfomance in Supply Chain,” Int. J.
Logist. Manag., vol. 26, no. 2, pp. 356–380, 2015.
The 7th International Conference on Cyber and IT Service Management (CITSM 2019)

[8] K. Kokini, A. Banerjee, and T. A. Taylor, “Thermal fracture of Volume Of The Hippocampus At One And Seven Day-Old
interfaces in precracked thermal barrier coatings,” Mater. Sci. Chicks,” Neurosci. Res. Commun., vol. 31, no. 1, pp. 29–38, 2002.
Eng., vol. 323, no. 1–2, pp. 70–82, 2002. [13] R. Hart and A. M. Kuo, “Meeting Health Care Research Needs in
[9] P. Alpar and M. Schulz, “Self-Service Business Intelligence,” Bus. a Kimball Integrated Data Warehouse,” DSAA, pp. 697–705,
Inf. Syst. Eng., vol. 58, no. 2, pp. 151–155, 2016. 2016.
[10] C. L. P. Chen and C. Zhang, “Data-intensive applications , [14] K. Dehdouh, F. Bentayeb, O. Boussaid, and N. Kabachi, “Using
challenges , techniques and technologies : A survey on Big Data,” the column oriented NoSQL model for implementing big data
Inf. Sci. (Ny)., vol. 275, pp. 314–347, 2014. warehouses,” Int. Conf. Parallel Distrib. Process. Tech. Appl., pp.
[11] H. Leslie, H. Spits, and R. Randriatoamanana, “Datawarehouser : 469–475, 2015.
A Data Warehouse artist who have,” TENCON, pp. 2205–2208, [15] J. George, B. V. Kumar, and V. S. Kumar, “Data Warehouse
2016. Design Considerations for a Healthcare Business Intelligence
[12] B. Sahin, H. Aslan, S. Canan, S. Kaplan, B. Unal, and E. Odaci, System,” World Congr. Eng., no. 1, pp. 4–7, 2015.
“A Stereological Estimation Of Total Neuron Number And