SPATIAL INTERPOLATION USING FAST INCREMENTAL GAUSSIAN MIXTURE NETWORK MODEL

• Main Author: HUTARI GANI, PRATI
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/23699
• Institution: Institut Teknologi Bandung

Completeness of information has an important role to determine the quality of <br /> <br /> information in the organization. Due to their limitations, most organizations only <br /> <br /> get some sample points in representing large geographical areas. An unobserved <br /> <br /> sample point affects in obtaining accurate information. An unobserved sample <br /> <br /> point can be obtained by interpolating the point known in the environment. In <br /> <br /> general, geostatistical methods are used on spatial interpolation. The disadvantage <br /> <br /> of geostatistical methods is to require variograms as inputs where these <br /> <br /> variograms require expert knowledge. This variogram is an obstacle for <br /> <br /> companies that do not have geostatistics experts. (Soares, Neto, & Roisenberg, <br /> <br /> 2016) uses the Incremental Gaussian Mixture Network (IGMN) model to solve <br /> <br /> organizational problems if they do not have geostatistics experts. (Pinto & Engel, <br /> <br /> 2015) developed the IGMN model by lowering the processing complexity and <br /> <br /> maintaining the quality of the results called Fast Incremental Gaussian Mixture <br /> <br /> Network (Fast IGMN) model. This Fast IGMN model has never been applied in <br /> <br /> completing spatial interpolation. Therefore, this study proposes the development <br /> <br /> of Fast IGMN model to handle spatial interpolation taking into account the <br /> <br /> necessary modifications. <br />