SHAPLEY Procedure

METHOD Statement

METHOD method-type <(options)>;

The METHOD statement specifies the method to use in the model explanations.

You can specify the following method-types:

HYPERSHAP

uses the hyperparameterized Shapley value estimation (HyperSHAP) method.

You can specify the following option:

DEPTH=number

specifies the number of subsets to use in approximating the Shapley values.

By default, DEPTH=1.

KERNELSHAP

uses the kernel Shapley value estimation (Kernel SHAP) method.

You can specify the following options:

BINWIDTH=number

specifies the bin width (relative to the standard deviation) to use in creating the binary encoding for interval variables.

By default, BINWIDTH=0.1.

INCLUDEMISSING

treats missing values of nominal variables as a valid level.

SAMPLESIZE=number

specifies the approximate number of observations to generate. The default is 500 times the number of variables that are listed in the INPUT statement or 1,000,000, whichever value is smaller.

SEED=number

specifies the seed value for pseudorandom number generation. If you do not specify a seed, or if you specify a value less than or equal to 0, the seed is generated by reading the time of day from the computer’s clock.

USERAWDATA

uses the raw reference data to calculate Shapley values and skip generating data. If you omit this option, the procedure generates data that have variable values from the variable distributions of the reference data table.

Last updated: August 06, 2026