shap4j

Java interface for the SHAP (SHapley Additive exPlanations) library for tree ensembles (TreeExplainer). Note that shap4j is not a pure Java port of SHAP. Rather, it utilizes JavaCPP to provide a Java-Native Interface (JNI) on top of the fast C++ implementation of TreeExplainer. In this sense, shap4j leverages the same underlying native code that powers the Python version of TreeExplainer, to ensure validity and efficiency.

Current supported platforms:

• macosx-x86_64
• linux-x86_64

Use cases

shap4j enables lean SHAP integration in JVM projects, i.e. a project can import shap4j as the sole dependency, without having to depend on heavier third-party tree ensemble libraries, e.g. xgboost4j.

Tree ensemble model conversion

In order for shap4j to explain a tree ensemble model, that model must be provided in a .shap4j data file, which can be generated from model dumps (pickle files) of XGBoost/LightGBM/CatBoost/sklearn using the companion Python library shap4j-data-converter.

Usage

Maven

<dependency>
  <groupId>io.github.xydrolase</groupId>
  <artifactId>shap4j-platform</artifactId>
  <version>0.0.2</version>
</dependency>

SBT

libraryDependencies += "io.github.xydrolase" % "shap4j-platform" % "0.0.2"

Example usage

package examples;

import java.nio.file.Files;
import java.io.File;
import shap4j.TreeExplainer;

class ExampleApp {
    public static void main(String[] args) throws Exception {
        byte[] data = Files.readAllBytes(new File("boston.shap4j").toPath());
        TreeExplainer explainer = new TreeExplainer(data);
        double[] x = {
                6.320e-03, 1.800e+01, 2.310e+00, 0.000e+00, 5.380e-01, 6.575e+00,
                6.520e+01, 4.090e+00, 1.000e+00, 2.960e+02, 1.530e+01, 3.969e+02,
                4.980e+00
        };
        double[] shapValues = explainer.shapValues(x, false);

        System.out.println("SHAP values: " + Arrays.toString(shapValues));
    }
}

The data file boston.shap4j can be found here. Or, you can follow the examples in shap4j-data-converter to generate it yourself.

API Docs

• TreeExplainer