semiArtificial

Contains methods to generate and evaluate semi-artificial data sets. Based on a given data set different methods learn data properties using machine learning algorithms and generate new data with the same properties. The package currently includes the following data generators: i) a RBF network based generator using rbfDDA() from package 'RSNNS', ii) a Random Forest based generator for both classification and regression problems iii) a density forest based generator for unsupervised data Data evaluation support tools include: a) single attribute based statistical evaluation: mean, median, standard deviation, skewness, kurtosis, medcouple, L/RMC, KS test, Hellinger distance b) evaluation based on clustering using Adjusted Rand Index (ARI) and FM c) evaluation based on classification performance with various learning models, e.g., random forests.