End members, end points and extragrades in numerical soil classification

摘要: Abstract Soil classification has progressed with the introduction of computers in mid 20th century to point where algorithms can be used organise soil information into clusters that correspond classes. Algorithms such as fuzzy-k means perform well, but biased by extreme data. Fuzzy-k extragrades was devised accommodate this problem estimating amount challenging and lead dubious classifications. The idea end members is discussed it concluded points, observations represent most parts continuum, are useful identification extragrades. We present discuss a new clustering algorithm, akromeson which identifies points given data set converts them pseudo clusters, then run concurrently semi-supervised algorithm. constructed synthetic order compare method It able correctly fix positions centroids, (which beyond capacity means), estimated were genuine extragrades, outperforming evaluated performance on from Edgeroi region New South Wales, Australia. algorithm identified an cluster periphery data, determined how use routinely find clusters. ability efficiently may provide added advantage pedologists generally stakeholders when they assessing land practices, especially regard areas exhibit properties require careful management, capable detecting.