Usability Analysis of Business Intelligence Tool Based Table Virtualization

ABSTRACT


INTRODUCTION
The Utilization of BI has been agreed that it increases the productivity of an organization from various sides [1]. The problem is that the resources needed for procurement, development and implementation in a normative context are quite significant. For example, in physical procurement and Data Warehouse implementation process, these kinds of process are required in BI development [2], [3]. In fact, this was difficult to realize by some organizations/companies primarily at the small-mid level company. An alternative solution offered by many cloud-based BI Tool developers is the next choice. The presence of products that offer the effectiveness and efficiency of BI development process is still difficult to be implemented entirely for several reasons.
Based on the above ideas, writer proposed alternative solutions of development of BI tool, especially for a small-mid level organization. This solution is to use an application which is relatively cheap and easy to use. The method of application selected in this study was then combined with the utilization of data table virtualization. Discussion on this study is directed to answer the problem of how the development of BI Tool is done and how far the performance of the results (output) is reviewed based on user perception on aspects of usability.
The study was conducted on one company at Small-Mid Enterprise (SME) level to obtain an important indicator that can be used as parameters in building an adaptive BI Tool for related companies. The obtained indicators are intended as materials to define user and system requirements that are then integrated into the standards and procedures for developing the BI Tool. Development is implemented and limited to the summary and detail report of the company's CRM database without going through the physical data warehouse process. The results of developed BI Tool then discussed, evaluated and analyzed to obtain conclusions in this study.  Simon (1997) categorizes BI Tools into 4 (four) forms, namely Basic Reporting tools, Online Analytical Processing, Data Mining, and Executive Information System. These tools provide information that focuses on executive-level-insight (information presented for executive needs) which operated with dashboards and other forms of display [4].
The related terms; ad-hoc query [5] and analysis tool are analytical tools included in BI Tools designed to enable users to find what-if business information without technical assistance from IT organizations [6]. Tools like this are usually in the form of a view with navigation that allows users to select the data relevant to the condition of the desired business report. This BI tool is also referred to as Business Intelligence Front-end [7].

Data and Table virtualization as SQL View
BI Tools uses data as material to be displayed in the form of information. Sherman [5] calls out forms of data that could be sources of reports and analysis are; OLAP databases, massively parallel processing (MPP) databases, data virtualization, In-database analytics, In-memory analytics, cloud-based BI and NoSQL databases. For sources processing, Johnson rejects the notion that physical Data Warehouse is the only enterprise solution in this issue [8]. Virtualization is one of alternative solution. It further explained that the Virtual Data Warehouse is identical to the federated model implementation process.
Strengthening the opinion above, in accordance with the white paper released by Pentaho states that MySQL (MariaDB) has convincing evidence that it can be used directly for data analysis (BI) primarily through the View function (Virtual Table) which features in MySQL version 5 [9]. Thus, MySQL with view function can be used for BI Tools data source because it admits as data virtualization and becomes one form of VDW implementation [10]. Data virtualization using view on MySQL/MariaDB also becomes a technology solution for small and medium-sized enterprise (SMEs), a solution to benefit in an enterprise context [11].

BI Tools Development by PRM
PHP Report Maker (PRM) is a software that can produce PHP-based Web Scripts to display reports from various database sources such as MariaDB, PostgreSQL, Microsoft Access, Microsoft SQL Server and Oracle database. It is a class of software that is very easy to use but has rich features that can be equivalent to software that is widely known by BI Tool developers. Some parties have used it as; dashboard development to monitor academic information at the university [12], build support systems with fuzzy methods [13]. Reporting inventory tools [14]. While Scriptcase is a software that specifically has the same function as PRM in building BI Tools as promoted on its official portal.

METHODOLOGY 3.1. Development Framework
To facilitate the understanding and the flow of analysis in this paper, we draw the order of framework as follows:  The picture above shows the design concept of BI Tool that will be built and analyzed in this research. This framework also explains the methodological steps in the study. To obtain data for analysis, the study was conducted in several stages. Starting from the initial data collection stage, the design and development stage, the publication stage and ending with access by the user. The data obtained through the process of BI Tool development, the results of generated output compares with the results derived from competitors output. The researcher uses Scriptcase [15] as comparison software in this study. In the last, the survey and post-test will be held to obtain user perception data over BI Tool interface generated.

Usability Analysis
The BI Tool performance concerning the use of conformity with user requirements and userusability aspects will be tested based on the 2 (two) ISO/IEC 25010 quality standard, both are Functional and Usability [16] The researcher concludes that taking two quality standards has represented required aspects of perceived analysis for the initial development of an application. According to [17] it only takes usability variables to know the quality and performance of software that is temporarily developed based on user experience. To measure both, the required measuring tools that have been tested as follows:  Analysis based on Developer Evaluation/Researcher, performed by heuristics evaluation/review method to evaluate the usability of an app's display by an expert pursuant to rules or policies enacted [18], with the following formula: A calculation pattern designed by David Travis was used to summarize the scores the analysis [20] to assist the researcher in completing the calculations of this evaluation.  Analysis based User Experience data, performed using Usability Metric for User Experience (UMUX) Lite. UMUX-Lite is a tool provided to measure user perceptions about the usefulness of an application/software based on user experience according to two raw variables. The formula applies to UMUX-Lite as follow: UMUXLITE=0.65(item1+item2-2) (100/12)+22.9 [21] 4. RESULT AND ANALYSIS

Requirements
The following table is the standard description used for synchronization requirement which is then initialized with the code for easy subsequent classification: The functionality presented provides correct results with the solid output precision level FS03 Existing functionality facilitates certain work to be done by the user.
The classification table above used as the basis for the preparation of the user and system requirements listing. Both requirements act as the guidelines in the development process. Also, resulting list of parameters used for the analysis. The detailed user and system requirements table are listed in attachments.

BI Data Modelling (Star Scheme)
The study uses star scheme as data modelling to produce fact tables. Meanwhile, view (as virtual table) is a fact table obtained from the relation of several dimension tables contained in the database. Data model adapted to the system requirement, a simple example can be described below: The fact table in the example figure above is constructed from several dimensional tables in view form. Due to its virtual nature, the fact table that is logically stored in the database with the name of the "viewTransactionServices". Thus, the table does not display its physical form. However, if it is compared to BI data modelling by involving logical table then this fact table will be tangible as in Figure 3 follows:  Table "viewTransaksiLayanan" (Simulated using Custom View Feature in PRM) Based on our experiment performed using PRM's custom view facility with the data model as shown in Figure 3 found that the data output is 100% like the table which produced the model in the logical model ( Figure 2). Furthermore, the results of the table virtualization created following the requirements used in PRM and SC for the further process of the BI Tool development.

Process of BI Tool Development
BI Tool development process using PRM and SC is relatively easy to do. This is because all activities can be done by utilizing the features that are available and are directed specifically to produce the report display in the form of tables and graphs. The summary of the process can be seen in the comparative process table as follows: The table above shows the summary of overall process of BI Tool developments using two software which proposed in this study (PRM and SC). In general, both have the same process. However, the facts show that the PRM process in some facets is easier to do with relatively better results. For example, in PRM it was found that the integrity of one process and other processes was very strong. The integrity of this process allows development is running more systematically. Where systematic process will greatly support the realization of a good SDLC as well.

Output PRM and SC
The result is a collection of PHP scripts and other files generated through the process of generating and compiling. The file generates several web pages that have certain functions and features described as follows: In general, the output has been able to meet the needs of the requirement listed in this study. Specifically, several features and forms distinguish between the results obtained both as described in Table 3.

Evaluation and Analysis a. Heuristics Evaluation
Here is a graph showing the difference and the deviation between the scores came by the BI tool developed using PRM and Scriptcase. The scores taken from developer perception using heuristics evaluation method. The figure shows that the PRM output is higher in majority aspects. There is five output usability have higher advantages compared to SC. While SC only excels on one aspect that is the aspect with code USR02. The evaluation results also state that there is a similarity of usability score in both outputs, i.e. on aspects with code US04SR.

b. Perceived Usability analysis (UMUX-Lite)
The matrix applicable to UMUX-Lite is between 22.9 -87.9 where the lowest score is the best result with a median of 55.4. Based on this standard we can describe the comparison of output (output) of PRM and Scriptcase according to the user perception as follows: Figure 5. Perceive Usability Comparison to PRM and SC Output based on UX from Users The figure above explained that all users give their perception for output (BI Tool) produced by PRM better than output by Scriptcase. However, with the fact that all opinions under the median, it can be stated that both of this software has been stated to have a good usability level to be used in building the BI Tool. In other words, the results of the BI Tool both built using PRM and Scriptcase have met the user usability perceptions and deserve to be used for BI customer service needs in this company.

CONCLUSION
Subjective considerations including the limitations of infrastructure and resources owned by the company become a reason for building alternative BI Tool development in this study. The development makes use of relatively easy-to-operate and affordable software combined with table virtualization for ETL utilization. The process is done comparable with the use of similar software for building Adhoc Report, as well as the output it produces. The output consisting of summary and detail report in both table and graphic form (crosstab report and serial chart) that provide visualization of data not yet found on company CRM system.
Based on the evaluation and analysis of user perceptions, found that PRM output has a higher usability level than Scriptcase. Also, all the user state that the output PRM is more declared more suitable for