Spatial analysis from macroscale to microscale based on mutual information
Editorial: Comité organizador del XXX Congreso Nacional de Estadística e Investigación Operativa y IV Jornadas de Estadística Pública
ISBN: 978-84-690-7249-3
Año de publicación: 2007
Congreso: Congreso Nacional de Estadística e Investigación Operativa (30. 2007. Valladolid)
Tipo: Aportación congreso
Resumen
Mutual information between a process of interest and an observed process is measured in terms of the wavelet transform. Blocks of wavelet coeffcients are considered to compute conditional entropy. Different block designs are formulated, with their performances depending on the local and global characteristics of both the interest and observed processes. The size of the blocks can also change with the scale, that is, from low-resolution (macro-scale) to high resolution (micro-scale) levels. In practice, the dispersion of the region of interest, the range of spatial dependence, and the local variability of the underlying models, including intensity of the observation noise, determine the optimal design of the blocks, as well as the thresholding rule for discrimination of useful information at different scales.