NB. When scrolling parameter window with mouse wheel, it can happen that the cursor stops at a slider widget. In this case, continuing to roll the wheel will not scroll the window but will move the slider and change its value which could be unintentional. |
'Model with fixed k (no)' is a slider for setting the fixed number of bins (k) in a histogram (range 10-115). The only model is learned based on a histogram with the chosen k.
'Model with varying k and the lowest BIC (yes) ' is a slider for setting varying binning parameter k. The default is a fixed k= 25 with a half-length of vector k =3 (checkbox for varying k is ticked). This setting allows learning of multiple models with a vector of k {10, 15, 20, 25, 30, 35, 40}. Software selects the model with the lowest BIC.
'Threshold for fusing too close Gaussians' is a slider for tuning a merge of Gaussian distributions that are too close to each other (range 0-1; default 0.5)
'Threshold for vanishing Gaussians' is a slider for tuning a disappearance of Gaussian distributions with too low number of values; (range 0-10; default 1.0)
Histogram colors
Edge
Panels
Gaussian colors
Total of Gaussian densities
Reference class '0'
Low class '-1'
Lowest class '-N'
Up class '+1'
Highest class '+N'
'Area under Gaussians (0=transparent, 1=opaque)' slider; (range 0-1; defaults 0.5)
'Line width for Gaussians' defaults' slider (range 0-10; default 2.0)
probabilistic
cutoff
s.cutoff
'Defaults' resets all the parameters to default settings
'Learn model' fits probabilistic model on the test data which are presented as a histogram. Because information about the number of bins and their boundaries is not inherent to the data, but will influence a model, it is advisable to apply varying binning parameters to the test data. Each newly built model is visualized by clicking on "Learn model" button.
'Draw plots' allows to re-draw customized histograms and plots
'Draw heatmaps' constructs classification heatmaps other than by default