PaseStub

Java client stub for a PASE server.

Code Example

Take a look at the code example section of the PASE repository. The same operations can be executed using a PaseInstance:

    PaseInstance instance = new PaseInstance("localhost:5000"); // specify host

    Map<String, Object> parameters = new HashMap<String, Object>();
    parameters.put("normalize", true);
    instance.create("sklearn.linear_model.LinearRegression", parameters);

    parameters.clear();
    double[][] X = {{0,0}, {1,1}, {2,2}};
    parameters.put("X", X);
    double [] y =  {0,1,2};
    parameters.put("y", y);
    instance.callFunction("fit", parameters);

    // You will have to know the structure of the return value:
    ArrayList<Double> coef_ = (ArrayList<Double>) instance.getAttribute("coef_");
    System.out.println("coef_: " + coef_); // coef_: [0.5,  0.5] 

    parameters.clear();
    double[][] X2 = {{0.5, 1}, {1, 0.5}};
    parameters.put("X", X2);
    ArrayList<Double> predictions = (ArrayList<Double>) instance.callFunction("predict", parameters);
    System.out.println("predictions: " + predictions); // predictions: [0.75,  0.75] 

Installation

Clone this repository. Use the makefile in the build/ folder to build this project:

cd build/
make

pasestub.jar will afterwards lie in the root folder and the resources folder will contain all the dependecies needed to use this stub. You can alternatively then use maven to build this project too.