Land suitability evaluation is an important tool to achieve sustainable land use but this approach requires some time-consuming and costly practices. One of the acknowledged techniques to minimize these practices is feature selection. Feature selection is an effective method to reduce the dimensional aspect of data and it involves searching possible combinations of attributes in the data to find which subset of features predict the best results. The main purpose of this study is to determine the most efficient method and the most effective factors to determine land suitability in the Shavoor plain, in the southwest of Iran. Different Fuzzy membership functions obtained from the literature were employed and the weights for each parameter were calculated according to an Analytic Hierarchy Process (AHP) that relies on pair wise comparisons. Random Search, Rank search, Best Search and Genetic methods were used. Random Search method selected EC, ESP and Texture for land suitability mapping, Best search and Genetic methods selected EC, Gypsum and Texture and Rank Search method selected EC, CaCO3, pH, Gypsum and Texture. The Kappa coefficient was used for comparison, which was 0.82, 80, 0.79 and 0.79 for Random Search, Rank search, Best Search and Genetic methods respectively. Results show that there is a strong agreement between fuzzy and Random Search methods to determine the land suitability and random search are most suitable for land suitability classification in the study area.
|Publication status||Published - 2012|
|Event||The 8th International Symposium Agro Environ, Wageningen, the Netherlands - |
Duration: 1 May 2012 → 4 May 2012
|Conference||The 8th International Symposium Agro Environ, Wageningen, the Netherlands|
|Period||1/05/12 → 4/05/12|
Mokarram, M., Hamzeh, S., Haratian Nejady, A., Bartholomeus, H., & Bregt, A. K. (2012). Using feature selection method to determine the most effective factors for land suitability mapping as compare with fuzzy method. Abstract from The 8th International Symposium Agro Environ, Wageningen, the Netherlands, .