Examples with NFSFileVec water.fvec.NFSFileVec used on opensource projects

Example 26 with NFSFileVec

use of water.fvec.NFSFileVec in project h2o-3 by h2oai.

the class GLMBasicTestRegression method setup.

@BeforeClass
public static void setup() throws IOException {
    stall_till_cloudsize(1);
    File f = getFile("smalldata/glm_test/cancar_logIn.csv");
    assert f.exists();
    NFSFileVec nfs = NFSFileVec.make(f);
    Key outputKey = Key.make("prostate_cat_train.hex");
    _canCarTrain = ParseDataset.parse(outputKey, nfs._key);
    _canCarTrain.add("Merit", (_merit = _canCarTrain.remove("Merit")).toCategoricalVec());
    _canCarTrain.add("Class", (_class = _canCarTrain.remove("Class")).toCategoricalVec());
    DKV.put(_canCarTrain._key, _canCarTrain);
    f = getFile("smalldata/glm_test/earinf.txt");
    nfs = NFSFileVec.make(f);
    outputKey = Key.make("earinf.hex");
    _earinf = ParseDataset.parse(outputKey, nfs._key);
    DKV.put(_earinf._key, _earinf);
    f = getFile("smalldata/glm_test/weighted.csv");
    nfs = NFSFileVec.make(f);
    outputKey = Key.make("weighted.hex");
    _weighted = ParseDataset.parse(outputKey, nfs._key);
    DKV.put(_weighted._key, _weighted);
    f = getFile("smalldata/glm_test/upsampled.csv");
    nfs = NFSFileVec.make(f);
    outputKey = Key.make("upsampled.hex");
    _upsampled = ParseDataset.parse(outputKey, nfs._key);
    DKV.put(_upsampled._key, _upsampled);
    _prostateTrain = parse_test_file("smalldata/glm_test/prostate_cat_train.csv");
    _airlines = parse_test_file("smalldata/airlines/AirlinesTrain.csv.zip");
    Vec v = _airlines.remove("IsDepDelayed");
    Vec v2 = v.makeCopy(null);
    _airlines.add("IsDepDelayed", v2);
    v.remove();
    DKV.put(_airlines._key, _airlines);
    //    System.out.println("made copy of vec " + v._key + " -> " + v2._key + ", in DKV? src =" + ((DKV.get(v._key) != null)) + ", dst = " + (DKV.get(v2._key) != null));
    _airlinesMM = parse_test_file(Key.make("AirlinesMM"), "smalldata/airlines/AirlinesTrainMM.csv.zip");
    v = _airlinesMM.remove("IsDepDelayed");
    _airlinesMM.add("IsDepDelayed", v.makeCopy(null));
    v.remove();
    DKV.put(_airlinesMM._key, _airlinesMM);
}

Example 27 with NFSFileVec

use of water.fvec.NFSFileVec in project h2o-3 by h2oai.

the class KMeansTest method testCovtype.

// "datasets directory not always available"
@Test
@Ignore
public void testCovtype() {
    Frame fr = null;
    try {
        File f = FileUtils.locateFile("../datasets/UCI/UCI-large/covtype/covtype.data");
        // Ignore if large file not found
        if (f == null)
            return;
        NFSFileVec nfs = NFSFileVec.make(f);
        fr = ParseDataset.parse(Key.make(), nfs._key);
        KMeansModel.KMeansParameters parms = new KMeansModel.KMeansParameters();
        parms._train = fr._key;
        parms._k = 7;
        parms._standardize = true;
        parms._max_iterations = 100;
        parms._init = KMeans.Initialization.Random;
        for (int i = 0; i < 10; i++) doSeed(parms, System.nanoTime()).delete();
    } finally {
        if (fr != null)
            fr.delete();
    }
}