# SCOUTerRpack

## SCOUTer package in R

Simulating anomalous data is an extremely common procedure. However,
very little attention is paid to this step and it is usually defined ad
hoc, existing a lack of standard. In this package, a new framework to
simulate outliers directly controlling their outlying properties has
been proposed. This framework offers the possibility of generating data
sets with *all type of desired properties*, given that the user
can control the pair of statistics that essentially define outliers: the
Squared Prediction Error ( *SPE* ) and the
Hotelling-T^{2} ( *T*^{2} ). These metrics
evaluate in a complementary way how far is an observation from the
majority of a data set. Since Given an observation with initial values
for the statistics, a PCA model and target values for the statistics,
our simulation method drifts the observation in a direction that shifts
the initial *SPE* and the *T*^{2} until reaching
their target values.