3rd NICTE IOP Publishing

IOP Conf. Series: Materials Science and Engineering 725 (2020) 012099 doi:10.1088/1757-899X/725/1/012099

Data warehouse using Kimball approach in computer maniac

Kelvin Salim*, Lorita Damayanti, Marisya Puspita, Steven Liujaya, Abba
Suganda Girsang
Computer Science Department, BINUS Graduate Program-Master of Computer
Science, Bina Nusantara University, Jakarta, Indonesia 11480

*kelvin.salim@binus.ac.id

Abstract. It is difficult for small businesses to evolve without utilizing an

adequate system which is able to manage information within the business entity
efficiently. The reporting system commonly used utilizes query which consumes a
large amount of time in order to generate results fulfilling requirements set by the
user. In order to simplify and speed up reporting in digital world and eventually
results in improvement of business, the business entity needs a system which can
address problems quickly and easily. In this paper, data warehouse system which
could simplify and speed up the reporting provisions that is needed by the
business without having to modify the existing query being implemented. With
Kimball approaches, the data to be reported could be customized in various forms
in accordance with the requirements set by the entity.

1. Introduction
Computer Maniac manages its transactions as sales and purchase transactions. Purchase
transaction is defined as a transaction that takes place when staff restocks a product from
supplier. Sales transaction is defined as a transaction that occurs when customer buys a
product being handled by the staff. After all processes related to restocking of goods and sales
of goods are complete, the related transaction is then inputted into the system. After goods are
received from the supplier by warehouse staff, the type and quantity of the goods, as well as
all other related data pertaining to the purchase are then inputted to the system. Sales
transactions are recorded as sales are made through POS system.
In order to have the data which are more structured and easier to be analyzed, a system
which can analyze data quickly and efficiently is required. It is possible to use a query to
retrieve the data; however it might consume a large amount of time in its process of analyzing
the data. For the growing business, the amount of data to be analyzed will increase as time
passes and more data is inputted. Implementing a data warehouse system is a way to make the
data analysis becomes more efficient.
In this paper, a method to implement the data warehouse [1] for better analysis and
creating report within the company requirement is needed. The method should be easy to
understand and efficient. The Kimball life cycle [1], [2] method is chosen as it uses a bottom-
up approach which is faster to implement.

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
3rd NICTE IOP Publishing
IOP Conf. Series: Materials Science and Engineering 725 (2020) 012099 doi:10.1088/1757-899X/725/1/012099

2. Related Work
1. Data Warehouse
Data warehouse [1], [3] is a technology that aims at enabling the decision maker to make
better and faster decisions [4]. Data Warehouse is a data structure that is optimized for
distribution, mass storage and complex query processing [3]. In order to build a Data
Warehouse [1], [3], it is required to run ETL tools which has three tasks: (1) extracting data
from different data sources, (2) transforming data, (3) loading transformed data to data
warehouse. There are two components in ETL tools, one component extracts raw data from
different data sources (flat files, excel or csv files, web services, relational tables) and another
component loads the data into the staging database, then cleans and transforms the extracted
raw data and loads it into facts and dimension tables [5]. There are two phases involved in the
extraction process: (1) initial extraction, (2) changed data extraction. The extraction process is
only executed once after building data warehouse in order to fill it up with a huge amount of
data from sources. In transforming process, all data are being cleaned and confirmed so that
the data gained is correct, complete, consistent and unambiguous. The process includes data
cleaning, transforming and integration. In loading process, all data that has been transformed
will be loaded to multidimensional structure.

2. Business Intelligence
Business Intelligence (BI) is defined as the transformation of information into knowledge
and it has the ability to provide the right information to the right user at the right time to
support the decision-making process [6]. Business intelligence systems are very complex and
expensive to design and implement [7]. The complexity and importance of BI system
development necessitates a critical approach to successfully develop technically appropriate
as well as usable (people-oriented) BI systems that meet user needs [7]. Some approach of
Business Intelligence includes the Kimball lifecycle approach [2], Inmon’s Corporate
Information Factory [8] and Linstedts’ data vault model [9]. When designing a new BI
system, business users often restrict themselves within the performance limitations of their
current (known) systems; hence, they only utilize current information and fail to explore new
and improved key performance indicators that can enhance their decisions [7].

3. Proposed Method
In this paper, Kimball methodology that has nine steps for designing the data warehouse is
used:
Choose Business Process and Analysis. In this step, the required tables which
contain the data are chosen and transformed into the dimension tables. In this case there are
eight tables such as Product, ProductType, Brand, Staff, Customer, Supplier, PurchaseDetail
and SalesDetail table.
Choose the Granularity. In this step, the granularity or the relationship between data
and information within transaction table and fact table is analyzed. In this case, the grain is
number of sales quantities within a year based on customers region.
Create Dimension Tables. In this step, the dimension tables that are related to the
fact tables are starting to be created. In this case seven dimension tables such as
CustomerDim, StaffDim, ProductDim, ProductDim, ProductTypeDim, BrandDim, DateDim
and SupplierDim are created.
Create Fact Tables. In this step, fact tables which contain measurable data are
created. In this case there are 2 fact tables such as SalesFact, PurchaseFact tables.

2
3rd NICTE IOP Publishing
IOP Conf. Series: Materials Science and Engineering 725 (2020) 012099 doi:10.1088/1757-899X/725/1/012099

Storing Pre-calculation in the Fact Table. In this step, after the fact table is formed,
the function of each field that serves as a measurement field are determined.
Rounding out the dimension table. In this step, the dimension table to represent the
attribute hierarchy to ease the analysis is determined.
Choosing the duration of the database. In this step, the duration for the data that
will be presented is chosen so that the data behaviour is analyzable. In this case the data
chosen is of 1 years backward.
Slowly changing dimension. In this step, it is made sure that dimension tables are not
affected by transaction changes.
After all dimension tables and fact table are created, all dimension tables are connected
with the fact table. A technique that can be used for this is Star Scheme [1]. Star Scheme [1]
consists of a central data table, or fact table connected to one or more dimension tables. It is
called a star scheme because this model resembles a star, with points centered from the center.
The center of the star scheme consists of one or more fact tables and the points in the schema
are dimension tables that contain information on certain attributes in the fact table [10].

3
3rd NICTE IOP Publishing
IOP Conf. Series: Materials Science and Engineering 725 (2020) 012099 doi:10.1088/1757-899X/725/1/012099

Figure 1. Entity Relational Diagram.

4
3rd NICTE IOP Publishing
IOP Conf. Series: Materials Science and Engineering 725 (2020) 012099 doi:10.1088/1757-899X/725/1/012099

Figure 2. Star Scheme.

After creating all dimension tables and the fact tables, the data is stored using ETL process
tool like Pentaho Data Integration. The ETL process is shown on Figure 3 – 11.

Figure 3. Load BrandDim table.

Figure 4. Load CustomerDim table.

5
3rd NICTE IOP Publishing
IOP Conf. Series: Materials Science and Engineering 725 (2020) 012099 doi:10.1088/1757-899X/725/1/012099

Figure 5. Load ProductDim table.

Figure 6. Load ProductTypeDim table.

Figure 7. Load StaffDim table.

Figure 8. Load SupplierDim table.

Figure 9. Load DateDim table.

Figure 10. Load PurchaseFact table.

6
3rd NICTE IOP Publishing
IOP Conf. Series: Materials Science and Engineering 725 (2020) 012099 doi:10.1088/1757-899X/725/1/012099

Figure 11. Load SalesFact table.

4. Analysis Results
After the ETL process is finished, all the dimension tables and the fact tables are filled
with data which can be used as data warehouse source. In this section, the report dashboard is
made using Qlik Sense. The result is shown on Figure 12 – 14.

Figure 12. Chart of quantity sales for every customers region.

Figure 13. Round chart of product type that sold the most.

7
3rd NICTE IOP Publishing
IOP Conf. Series: Materials Science and Engineering 725 (2020) 012099 doi:10.1088/1757-899X/725/1/012099

Figure 14. Chart of brand that sold the most.

There also another tool to make such report such as Pentaho Report Designer. The result can
be seen on Figure 15.

Figure 15. Report result using Pentaho Report Design

8
3rd NICTE IOP Publishing
IOP Conf. Series: Materials Science and Engineering 725 (2020) 012099 doi:10.1088/1757-899X/725/1/012099

5. Conclusion
It is possible to make data easy to analyse and report faster to be made with a data
warehouse system without the need to type a long query. All related table are processed by
ETL process which processes all data needed in the data warehouse and separates all
unnecessary data. As the result, utilizing the process, it is possible for staff to prepare report
which fulfils company requirement within shorter period of time. In the future, can make it
easier and more complex reporting system than what already exists and make a reporting
system that is better and more efficient than now.

References
[1] G. M and S. Rizzi, “From Star Schemas to Big Data: 20+ Years of Data Warehouse
Research,” J. Biol. Chem., pp. 241–252, 2018.
[2] R. Kimball and M. Ross, The Kimball Group Reader: Relentlessly Practical Tools for
Data Warehousing and Business Intelligence. Indianapolis, Wiley, 2010.
[3] T. R. Sahama and P. R. Croll, “A Data Warehouse Architecture for Clinical Data
Warehousing,” Russ. J. Gen. Chem., vol. 78, no. 11, pp. 2214–2219, 2008.
[4] S. H. A. El-Sappagh, A. M. A. Hendawi, and A. H. El Bastawissy, “A proposed model
for data warehouse ETL processes,” J. King Saud Univ. - Comput. Inf. Sci., vol. 23, no.
2, pp. 91–104, 2011.
[5] S. Habte, K. Ouazzane, P. Patel, and S. Patel, “Generic Data Warehousing for
Consumer Electronics Retail Industry,” Int. J. Comput. Inf. Eng., vol. 11, no. 7, pp.
828–831, 2017.
[6] A. Brandao, E. Pereira, F. Portela, M. Santos, A. Abelha, and J. Machado, “Real-time
Business Intelligence platform to maternity care,” in 2014 IEEE Conference on
Biomedical Engineering and Sciences (IECBES), 2014, pp. 379–384.
[7] C. Venter and R. Goede, “Critical systems approach to business intelligence systems
development,” Proc. 59th Annu. Meet. ISSS-2015 Berlin Ger., pp. 1–18, 2016.
[8] W. H. Inmon, C. Imhoff, and R. Sousa, Corporate information factory. New York,
Wiley, 2001.
[9] D. Linstedt, “Method and system of data warehousing and building business
intelligence using a data storage model,” vol. 1, no. 19, 2002.
[10] S. M. Isa, E. C. Nugroho, D. Y. Gunarso, C. B. Process, K. C. Susena, and A. S.
Girsang, “Business Intelligence for Analyzing Department Unit Performance in
eProcurement System.”

9
Reproduced with permission of copyright owner. Further reproduction
prohibited without permission.