Membangun Kunci Identifikasi Asterinaceae Berhifopodia Menggunakan k-Nearest Neighbour KarakterMorfologi

Abstract

Each species of fungi is distinguished by a set of characters that are unique to that species. One of those are morphological characters. This study is aiming at setting up the interactive identification key of the Hyphopodiate Asterinaceae that consist of 37 species of Asterina, 1 species of Asterolibertia, and 4 species of Lembosia, using K-nearest neighbour method (KNN). KNN is one of classification methods that calculate similarity based on the distance between data training and data testing. A set of 116 morphological characters with two data types i.e 104 nominal data and 12 numeric data of 93 specimens were used to set up the models of the identification key. Euclidean distance was used for the numeric data type where as the Hamming distance was used for the nominal data type. The Euclidean distance and the Hamming distance will be combined by the aggregation function. The smallest value of aggregation indicated the degree of similarity between groups and the identity. Web-based interactive identification key of Hyphopodiate Asterinaceae was used as a primary source of information of the diversity of Hyphopodiate Asterinaceae in the world for mycologists and other users. Data from 42 species each were used as data testing. It was found that the accuracy of normalized data was higher than the unnormalized data, with values of 100% and 78.38% respectively. Therefore, this web-based identification was developed by using the normalized numeric data.