1. Introduction
All aspects in life need knowledge. Knowledge can be gotten from many sources. Before getting the knowledge, information must be read in proper way. Data owner can present their data in many forms. Sometimes, it can be difficult to extract information if data were presented in wrong ways. This thing can result in wrong knowledge and lead the business to the wrong path. Absolutely, business owners don’t want this thing happens. One critical sector that needs an easy and optimal way to extract information and knowledge is the poultry farm sector.
For a business like poultry farm, information may lead to the increment or decrement of profit. If the cage condition is not well maintained then failure may occur. Failure of a poultry farm can start from the breeder's error in cage maintenance and temperature regulation [1]. Failure in cage maintenance can cause disadvantages and reduced revenues.
The environmental factors such as air temperature, humidity, and ammonia level must be known by breeders. Ignorance of one of these can affect the growth of broilers. Management of poultry farms and breeders need tools to facilitate the monitoring processes for environmental factors in their poultry farms. The need for monitoring is also due to anomalies in the enclosure. The anomaly occurs because there are differences in the condition of one or several broilers with other broilers [2]. A large number of broiler chickens in each cycle period also makes breeders have difficulty in knowing causes of chicken death. Collection of data history can help the production and management of broiler breeding. Data which are shown at tools must be visualized in an appropriate way so that breeders can immediately know about proper strategies that must be taken.
One of the best ways to understand great data is with data visualization. Data visualization is a way to bring data more alive in the form of visual form to make them more efficient to understand and to be read [3-5]. With data visualization, the pattern in data can be extracted rather than just using the usual statistical methods. Visual forms, such as diagrams, symbols, and combination between them with texts, can represent environmental factors data to breeders and translate all of them in easiest way to be read. In this research, website will be developed to present data visualization in this domain case and usability testing will be conducted to measure how far the users can extract and understand about those visualizations.
The purpose of this study was to show the data visualization of environmental factor in poultry farm.
2. Research methodology
Research methodology, which was used to conduct this research, was referred to data visualization processes [6,7]. Based on that methodology, this research was conducted by using the methodology that can be seen at Figure 1.
Based on Figure 1, this research will be started by collect data from data source. After that, information needs would be identified for each user to filter data. Knowledge discovery was conducted as an enrichment process so that valuable information could be shown together with other visualization. To make this research more measurable, all of analyzed visualization were implemented in the form of software and usability will be held to ensure user’s understanding level about the visualizations.
3. Results and discussions
In this section, all stages in methodology will be discussed further. Explanations in this section is more about inputs, processes, and outputs of each stages. Some diagrams, figures, and tables will be presented to give more explanations. This section will be divided into five subsections same as methodology.
3.1. Acquire data from IoT sensors
Broiler data which will be used comes from two sources. First source is IoT sensors and second source is Excel documents that were filled by breeders manually. Data from IoT sensors was sent into Firebase so transforming data have happened here. The data structured were transformed into hierarchical structure with the intention of making data easier to be accessed. Data that were received by Firebase from the sensors are about how much chicken feed that were given per day and quality data of broilers.
Data from Excel documents were used as a comparator for IoT data. Sometimes, sensors can result in some errors and to fix this, breeders need a comparator. Data which will be used can be seen on Table 1.
3.2. Data filter based on user needs
Data filter in this research includes two stages, are parse and filter data. Parse is the process of classifying data that has been collected based on information needs to solve existing problems. While the filter is the process of mapping data based on information needs to be used in the next stage.
In this research, there are two groups of users. They are breeders and poultry farm owner. Breeders need information that specific about environmental factors in poultry while poultry farm owner need information that more specific about how success poultry farm that they owned. Based on that fact, information was mapped based on information needs per users as can be seen at Table 2.
3.3. Mining knowledge’s using DBSCAN
Based on Table 2, there are two information that need to be extracted into knowledge. To know about the existences of grid that have different condition with others, it can’t be done if not doing knowledge discovery [8,9]. Analysis of methods and algorithms (mine) is a process in which problem-solving is done using statistical calculations or certain algorithms [8-11]. Mapping methods and algorithms based on information needs can be seen on Table 3.
To find information about how broiler conditions and the environmental factors per day, descriptive statistics method was applied to data 1 and data 2 to find the average of each attribute. Contrary to that, DBSCAN was applied to data 2 to find outlier between grids in one cage. DBSCAN is a clustering method which be able to identifying noise in a group of data based on density between data points [12,13]. Some data from mobile sensor were chosen as sample data. Sample data come from January 1, 2018.
To determine the optimal eps value is to measure the distance between the center of the cluster with each data point used Euclidean Distance, then get the distance matrix [12,14]. In this example, the number of nearest neighbors (k) is 3 (3-Distance). Based on the k-distance results, they can be determined the optimal estimation of eps value for each data. Results of plots that have been sorted ascending are described by k-distance Graph at Figure 2. It can be seen that the optimal value of eps is in the range of number 71 with minPts worth 3 [13]. After the data is done for the implementation, cluster results were obtained. Results plot of clusters can be seen on Figure 2 and Figure 3.
3.4. Mapping visualization
Visualization aims to make data easy to read and understand [15]. In general, the presentation of data can be grouped into three parts: presentation in the form of text, presentation in the form of tables and presentation in the form of graphs or diagrams [16,17]. The diagram is used because it is easier to get the picture data as a visual tool for the layman [18]. The result of data visualization mapping can be seen on Table 4.
3.4.1. Symbol. The needs for this visualization is to show the condition of broilers and their environments from each grid. The conditions of broilers and the environment are described by symbols that represent the condition. Explanation of symbols used can be seen at Table 5. Results Form of visualization that is used these symbols can be seen at Figure 4.
3.4.2. Heat map. The needs for this visualization is to find out whether there is a broiler grid that is different from other broiler grids. This information is obtained from the calculation of DBSCAN clustering algorithm that has been done before. The results of cluster visualization can be seen at Figure 5.
3.5. Implementation and usability testing
In this last stage, a website to visualize information in this research was built. This is meant for users to try and be measured how understanding level of them. Usability testing was conducted to ensure that visualization in website meets the research objectives [19].
Scenario testing usability testing by defining the tasks that will be ordered to system testers, namely the cage officer. A defined task consists of two types: poor task and better task. The example of test scenarios can be seen at Table 6.
From the results of usability testing, the results of the effectiveness level are 100%. From these results can be concluded that the testing of broiler data visualization system has met the expected goals. The cage officer can do the job well provided and understand the knowledge of each displayed data.
4. Conclusions
Based on the results of implementation and testing, it can be concluded that the visualization in the developed software can help the cage officer in monitoring the environmental conditions of the cage and the development of broiler and besides that, symbols are more effective to be understood by users in this domain than other visualization forms. However, there are still constraints faced in visualizing data in real-time. A constraint if the data captured by the sensor cannot be stored into the database or the constraint due to error reading the data from the sensor. Therefore, for the development of a data visualization system in the future, optimizing communication between sensors and database and optimizing of data which is served by services will need research further.
This article was originally published in IOP Conf. Series: Materials Science and Engineering 407 (2018) 012120. doi:10.1088/1757-899X/407/1/012120. This is an Open Access article used under the terms of the Creative Commons Attribution 3.0 License. 
