Manual feature selection enabling to exclude unnecessary features from the dataset.
Automatic feature selection and features ranking based on the method of correlation coefficients.
Automatic feature selections and features ranking using any classification models included in GhostMiner as “wrappers”.
Automatic feature selections taking advantage of the committee's idea and voting, where committee is composed of any from above selectors.
New classification multimodel - Transform & Classify (any combination of feature selection method and classification model from GhostMiner) enabling to build a model to be validated on full scope of features.
Sampling of dataset enabling to work with large volume of data.
Additional very effective semi-automatic model for classification based on the Support Vector Machine method.
New models for solving clustering problems using Dendrograms and Support Vector techniques.
New properties and visualization of decision tree such as the cost function, class distribution for each leaf, colour shadowing etc., allowing for the full control during the procedure of the tree construction.
Additional visualization and data transformation in terms of Principal Component Analysis.
Leave-one-out testing tool.
Extended Analyzer facilitating simultaneous classification of multivector input nd enabling better communication with Developer due to open-save project functions.
Improved and simplified data import/export module facilitating handling of Excel and Text/CSV formats in more user friendly and effective way.
More versatile project structure.
Improved handling of complex models.
Print functions and graphics export to many bitmap and vector formats including JPG, BMP, PDF, HTML, and PostScript.
General improvements of the data viewing, plotting, and presentation.