APPLICATION OF CLUSTERING GROUPS IN DETERMINING LAND SUITABILITY

The problem of the suitability of planting land for planting types of plants is still a lot of unresolved farmers. land at the beginning of land selection is very decisive. This is to determine whether the land is productive or not. in the future is the result that farmers will experience considerable losses if this is not estimated, therefore a model is needed to know and estimate accurately whether the land or land is suitable for planting crops or not. The purpose of this study is that farmers do not experience losses and losses on land when planting crops on available land. So that it can improve the quality of food in Aceh and can improve the welfare of farmers. The result of the research is that it can be found a Clustering model or grouping of plant stock types using the k-medoid model, finally the application of this clustering model is able to provide a solution for the Office. Based on calculations using the K-Medoids method obtained for land compliance in North Aceh Regency is Corn with an adjustment of 53.6%, Peanuts 17.9%, Soybeans 14.3%, Bulbs 10.7% and the suitability for rice is 3 ,6%.


Introduction
Information technology is growing rapidly like today's technology in agriculture. Currently, there are many discoveries that are considered to improve the quality of agriculture in terms of the agricultural sector. Constraints faced in the agricultural sector greatly affect land suitability, and this will affect food security in the area [1]. Determining the classification status of suitable plantings on land is one of the techniques used in data mining, where classification is an activity in extracting data and then predicting category labels for each data. Food is something that is intended as food or drink for human consumption [2]. Then the food is intended as food or drink for human consumption. [3]. Groups of farmers who cultivate crops must prepare and detail the suitability of the land according to the types of plants planted so that the yields of these types of food crops will produce satisfactory results and do not suffer losses. The selection of land at the beginning that is unproductive and inappropriate, then the harvest will experience considerable losses later. This requires a clustering model that makes it easier for farmers to see what plants are suitable and in accordance with soil conditions in the area.

Data Mining
Data mining is divided into several groups, namely Classification, namely classification and categories [4]. Next is machine learning to extract and identify useful information and knowledge compiled from various large databases [5]. Techniques that can be used in the data stored in the database [6]. Clustering is a way of finding interesting patterns from large amounts of data, and then storing that data in databases, data warehouses or information stores [7]. The stages or steps taken for the data mining process start from the selection of data obtained from the data source to the target data, the pre-processing stage to improve data quality. Automatic pattern discovery in databases or other data sources [8] [9]. One of the data mining techniques is the classification function. Classification technique is a learning technique used to predict the attribute value of the target category [10]. Then the data mining algorithm is used for the classification process of teak tree data from Perum Perhutani Kph Semarang [11]. Then clustering is used to group the data into several clusters so that the data in one cluster has a maximum level of similarity and the data between clusters has a minimum similarity [12]. Cluster is an unsupervised data mining method [13].

Algoritma K-Medoids
This algorithm uses objects in a set of objects to represent a cluster. These objects are selected to represent a cluster called medoids [14]. K-Medoid or PAM algorithm uses the object (medoid) as the center of the cluster for each cluster, while K-Means uses the mean as the center of the cluster [15]. Another advantage is that the results of the clustering process do not depend on the order of entry of the data set [16].

Land
Land use is included in the socio-cultural component because land use reflects the results of human activities on land and its status. Land evaluation is carried out through the process of assessing land resources for certain purposes using a tested approach. The results of the land evaluation will provide information and directions for land use as needed. The results of this evaluation will later provide information which is then used as a direction for land use to suit the needs [17]. Increasing the productivity and quality of agricultural products directly affects the income and welfare of farmers, and is at the core of agricultural economic development. One of the factors that support the trigger for increasing productivity and quality of agricultural products is the use of information technology in the agricultural sector [18].

Food
Food is also an unavoidable physical need, which in terms of anthropologist Melvile J. Herkovitas, is the primary determinants of survival for mankind [19]. Local food is something that is consumed by local people in accordance with local potential and wisdom. Processed food or processed food is food that is processed by certain methods. Food is an inevitable physical need, which in anthropologist Melvile J. Herkovitas' terms, is the primary determinants of survival for mankind. Processed food or processed food is food or drink that is processed in a certain way or method with or without additional ingredients [20].

Method
The research method used is descriptive method with a quantitative approach. Methods of collection with Field Research (Field Research), Interviews, literature study data collection sourced from manual books, literature compiled by experts to complete the data needed in research [21]. The research stages describe the research process as well as describe the research as a whole [22]. The research stages include data processing carried out according to the k-medoids clustering method from existing data from clustering clustering data. The results of the clustering implementation process are carried out using the K-Medoids method. The design of this research went through several stages, namely analyzing through field research to obtain information and data [23], calculating all data using the kmedoids method, then designing the system, designing context diagrams, data flow diagrams (DFD) and implementing the system to be built [24]. The following scheme of the research system is as follows:

Analysis System
Problems in the use of suitable land for planting crops are needed at this time. Directions in the use of land to plant types of plants that are in accordance with the land to be planted, which then there are still many farmers who do not know and take into account exactly whether the land or land is suitable for planting crops. With the k-medoids clustering model, this system can improve the food quality of North Aceh and can improve the welfare of farmers. Implementation or implementation can be done by clustering using the K-Medoids method.

System planning
The following is a context diagram that describes the processes and work flow of the system as a whole.

K-Medoids Method Manual Calculation
In this manual calculation, there are 10 sample data that will be calculated using the K-Medoids Method, which aims to get the results of the appropriate land suitability for plants in North Aceh district. The sample data from each sub-district are as follows:

Determining the Data Centroid
To perform calculations, you must first determine the centroid and initialize the cluster name as follows:

Normalization Calculation
Determination of the normalization value, here must first find the maximum value and drink from the sample data above, which is as follows: After obtaining the maximum and minimum values from the data of all existing data, then calculating the normalization for each data using the normalization formula, following are the results of the manual calculation of data normalization:

Initial Cluster Value
To perform calculations using K-medoids, first determine k or the desired number of clusters, in this study using 5 clusters so that to perform calculations the initial medoids were randomly selected from the data for each sub-district as many as k clusters as the cluster center. Furthermore, all the data is calculated the distance, the next step is to group the data according to the cluster, the cluster group of a data is calculated from the closest distance from the data to a cluster so that the cluster results are obtained as shown in the table below:   After the results of the 1st and 2nd iterations are obtained, the distance value is obtained, then calculate the total deviation (S) by finding the difference between the old total cost minus (-) the new total cost. With the provision that if S < 0, then the object value is exchanged by determining a new medoid. The process is repeated from step 4 to step 6. S = old total costnew total cost = 7,850341236 -7,194100262 = 0,656240974 Because the final result of the total deviation is greater than zero ( S> 0 ). So this test is stopped in the 2nd iteration. The result of the last iteration using the initial medoids 2 will be the clustering parameter. So that it can be seen the results of the clustering and the number of classes according to the cluster name as follows; From the results of the above calculations, the results obtained for the number of each class from each cluster are as follows: Table 9. Result Clustering K-Medoids

Implementation of K-Medoids Display Display 1. Data Normalization Results
The data normalization display in the application of K-Medoids and Borda Clustering Grouping Based on Plant Types in North Aceh Regency is as follows:

Medoids algorithm as cluster center
The display of the initial Medoids Algorithm as the center of the cluster in the Application of K-Medoids and Borda Clustering Grouping Based on Plant Types in North Aceh Regency is as follows:

Conclusion
The conclusions in the application of K-Medoids and Borda Clustering Grouping Based on Plant Types in North Aceh Regency are as follows: 1. The implementation of data mining in grouping soil types according to plants based on the k-medoid clustering method and the application of forecasting technology applications to be able to predict the types of commodities in plant species 2. Able to monitor the types of plant commodities in each area in the cluster using the kmedoid grouping model method in determining the types of plant commodities used for each regional need.