Missing value imputation with 20-nearest neighbors
2D plots of additive effects for visualizing nonlinearity
3D plots for visualizing interactions
Overall fit measures GoF
Fit measures structural model: R2
Fit measures measurement model: Crombachs Alpha, AEV, Composite Reliability
Path strenght measures (standarized and unstandardized): Average Simulated Effect (ASE), OEAD, Linear Path Coefficient, Absolute Maximum Effect
Factor Scores
Polinomial formula for every nonlinearity
Interaction strength effect IE
T-values and significance level for ASE, OEAD, LPC, IE
Confidence intervales 10% – 90%
Nominal variables as gender can be incorporated like every other MV
Estimates Total Effects by summarizing direct and indirect impacts
Options
Reflexive and Formative constructs usable
Invert scale of MV to improve readability of plots
Normalizing scale to 0-100% to improve readability of plots
Disable Bootstrapping and plottings
Over-weighting of rare cases.
Cross-validation and Hold-One-Out
Specify No. of interations
Specify NN size and size of Committee-of-Networks
Specify apriori probablility of paths
Linear Partial Least Squares
Modules
Second-order models
Hierarchical Bayes for calculation of individual OEAD (path strength)
CAA Competitive Advantage Analysis
TS time serial data analysis
Missing Imputation
Case weightning
Segment-wise modeling
2-Stage Least Square correction (… of endogenious variables by using instrumental exogenious variables. This enables true models although important variables that influence several endogenious variables are missing.)
Additional revolutionary features
Causal Direction Discovery: Evaluates bidirectional paths and potentially identifies true causal direction.
Universal Multi-Target Regression: Regression method that can handle fewer cases than input variables by regressing on multiple target variables simioutaniously.
