Oil Palm Plantation Land Suitability Classification using PCA-FCM

Abstract

Classification of land suitability is very important for oil palm plantation for oil palm and quality of oil palm yield. This classification depends on soil properties and climate characteristic of Pelalawan district, Riau province. Soil properties are combinations of soil characteristics which are known to occur in soils
and which are considered to be indicative of present or past soil-forming processes. In addition to soil properties, land suitability also is influenced by the climate in that area. Climate components that influence the growth of palm oil are the air temperature, rainfall and humidity. Furthermore, to classify land suitability will be classification algorithm that can be used for analysis of soil properties and climate characteristic. In this dissertation, the author will be used combining of the Principal Component Analysis (PCA) and Fuzzy Clustering Means (FCM) methods for classifying of land suitability. These techniques were applied to the collected soil and climate data and the achieved performance were joined and analysed. Based on experimental results, PCA and FCM were able to improve the performance of accuracy in classification. PCA – FCM can achieve 98.780% for training process and 97.345% for testing process when using six
variables. While comparative study, PCA-FCM can achieve 98.890% for training process and 98.655% for testing respectively. By applying PCA to classify land suitability, the accuracy of classification can be maintained with less of the number of variables.