STRATIFIED RANDOM SAMPLING FOR WATER AND NON-WATER REGION CLASSIFICATION USING PYTHON

摘要

The main purpose of the stratification is to provide a higher degree of relative efficiency by giving better cross-section of the population where the reliability of the accuracy is determined through the sample units and sample schemes. To reduce the high sampling variance, the division of population into subpopulation may be used, this study performed stratified random sampling using Arcpy Package to be used as sampling array to classify water and non-water region. Samples were tested in Bayes. Function, and Decision Tree algorithms. Result shows that NaTvebayes has 98.4452 Overall Accuracy with 0.6047 kappa. .148 has 07.9752 Overall Accuracy with 0.7209 kappa, and Sequential Minimal Optimization (SMO) has 98.7659 Overall Accuracy with 0.626 kappa values.