Aplikasi Model Artificial Neural Network Terintegrasi dengan Geographycal Information System untuk Evaluasi Kesesuaian Lahan Perkebunan Kakao

  • Hermantoro . Fakultas Teknologi Pertanian, Institut Pertanian Stiper
  • Rudiyanto . Fakultas Teknologi Pertanian, Institut Pertanian Bogor
  • Slamet Suprayogi Fakultas Geografi Universitas Gadjah Mada


Land evaluation for specific purpose in plantation sector become very important due to increasing the competition in land use and the development of plantation sector. Land evaluation produces information of land economic values for specific land use. The objective of the research is to develop land evaluation method for cocoa estate using integrated model Artificial Neural Network (ANN) and Geographical Information System (GIS). Back propagation ANN model were used to predict cocoa yield base on land qualities parameter. The result shows that the best of ANN model to predict cocoa yield have 15 input layer, 15 hidden layer, and 1 output layer. with the determination coefficient (r2) of 0.99 and Root Mean Square Error (RMSE) of 93.83 in the training process, otherwise in the testing found the r2of O. 76 and RMSE of 113.83. In verification stage the integrated model ofANN and GIS was used to evaluate land suitability of Wijayaarga Cocoa Plantation is seem accurate in predicting cocoa yield and easers to mapping the land suitability unit.

Keyword: ANN, GIS, Land Evaluation, Cocoa

Diterima: 04 Juni 2007; Disetujui: 18 September 2007