Improve Hybrid Particle Swarm Optimization and K-Means by Random Injection for Land Clustering of Potato Plants

Potatoes are a source of carbohydrates that have lower glucose levels than rice so it is safe for diabetics. potatoes are agribusiness superior products for the highlands because of their high economic value and have great potential in product diversification. Batu City is a potato producing center in East Java which is spread in the Sumber Brantas village, Tulungrejo village and its surroundings. Based on statistical data, the average productivity of potato crops in the city of stone is 75,860 Kw. Potato crop productivity is not optimal due to the varied variability and productivity of the soil. Based on these problems it is necessary to cluster the potato planting land based on the level of land suitability. Land clustering is grouped into 4 classes: very suitable (S1), quite appropriate (S2), appropriate (S3) and inappropriate (N). Used 11 Land suitability criteria, namely: average temperature, first month rainfall, second to third month rainfall, fourth month rainfall, air humidity, drainage, soil texture, effective depth, H2O pH, CEC and slope slope. K-Means is a simple clustering algorithm that has properties without direction or unsupervised. In complex problems, the K-Means method often gets a non-optimum solution (Local Optimum). In this study a new approach was used, namely hybrid particle swarm optimization and K-Means (KCPSO). KCPSO is implemented using the random injection concept to obtain better fitness values. Calculation of fitness value based on Silhouette Coefficient value. Based on the results of testing, the KCPSO architecture imported by random injection obtained the best fitness value. The clustering process using improve KCPSO compared to expert calculations produces an accuracy rate of 86%