True, the feature space is transformed using the method defined under the
zscore. The other available options are:
z-scoreThe standard zscore is calculated as z = (x – u) / s
minmaxscales and translates each feature individually such that it is in the range of 0 – 1.
maxabsscales and translates each feature individually such that the maximal absolute value of each feature will be 1.0. It does not shift/center the data and thus does not destroy any sparsity.
robustscales and translates each feature according to the Interquartile range. When the dataset contains outliers, the robust scaler often gives better results.
True, a power transformer is applied to make the data more normal / Gaussian-like. This is useful for modeling issues related to heteroscedasticity or other situations where normality is desired. The optimal parameter for stabilizing variance and minimizing skewness is estimated through maximum likelihood.
yeo-johnson. The other available option is
quantiletransformation. Both the transformation transforms the feature set to follow a Gaussian-like or normal distribution. Quantile transformer is non-linear and may distort linear correlations between variables measured at the same scale.
True, the target variable is transformed using the method defined in
transform_target_methodparameter. Target transformation is applied separately from feature transformations.
box-coxrequires input data to be strictly positive, while
yeo-johnsonsupports both positive and negative data. When
box-coxand target variable contains negative values, the method is internally forced to
yeo-johnsonto avoid any exceptions.