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

Example 11 with Chunk

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

the class AstCut method apply.

@Override
public ValFrame apply(Env env, Env.StackHelp stk, AstRoot[] asts) {
    Frame fr = stk.track(asts[1].exec(env)).getFrame();
    double[] cuts = check(asts[2]);
    Arrays.sort(cuts);
    String[] labels = check2(asts[3]);
    final boolean lowest = asts[4].exec(env).getNum() == 1;
    final boolean rite = asts[5].exec(env).getNum() == 1;
    // cap at 12
    final int digits = Math.min((int) asts[6].exec(env).getNum(), 12);
    if (fr.vecs().length != 1 || fr.vecs()[0].isCategorical())
        throw new IllegalArgumentException("First argument must be a numeric column vector");
    double fmin = fr.anyVec().min();
    double fmax = fr.anyVec().max();
    // c(0,10,100) -> 2 bins (0,10] U (10, 100]
    int nbins = cuts.length - 1;
    double width;
    if (nbins == 0) {
        if (cuts[0] < 2)
            throw new IllegalArgumentException("The number of cuts must be >= 2. Got: " + cuts[0]);
        // in this case, cut the vec into _cuts[0] many pieces of equal length
        nbins = (int) Math.floor(cuts[0]);
        width = (fmax - fmin) / nbins;
        cuts = new double[nbins];
        cuts[0] = fmin - 0.001 * (fmax - fmin);
        for (int i = 1; i < cuts.length; ++i) cuts[i] = (i == cuts.length - 1) ? (fmax + 0.001 * (fmax - fmin)) : (fmin + i * width);
    }
    // if(width == 0) throw new IllegalArgumentException("Data vector is constant!");
    if (labels != null && labels.length != nbins)
        throw new IllegalArgumentException("`labels` vector does not match the number of cuts.");
    // Construct domain names from _labels or bin intervals if _labels is null
    final double[] cutz = cuts;
    // first round _cuts to dig.lab decimals: example floor(2.676*100 + 0.5) / 100
    for (int i = 0; i < cuts.length; ++i) cuts[i] = Math.floor(cuts[i] * Math.pow(10, digits) + 0.5) / Math.pow(10, digits);
    String[][] domains = new String[1][nbins];
    if (labels == null) {
        domains[0][0] = (lowest ? "[" : left(rite)) + cuts[0] + "," + cuts[1] + rite(rite);
        for (int i = 1; i < (cuts.length - 1); ++i) domains[0][i] = left(rite) + cuts[i] + "," + cuts[i + 1] + rite(rite);
    } else
        domains[0] = labels;
    Frame fr2 = new MRTask() {

        @Override
        public void map(Chunk c, NewChunk nc) {
            int rows = c._len;
            for (int r = 0; r < rows; ++r) {
                double x = c.atd(r);
                if (Double.isNaN(x) || (lowest && x < cutz[0]) || (!lowest && (x < cutz[0] || MathUtils.equalsWithinOneSmallUlp(x, cutz[0]))) || (rite && x > cutz[cutz.length - 1]) || (!rite && (x > cutz[cutz.length - 1] || MathUtils.equalsWithinOneSmallUlp(x, cutz[cutz.length - 1]))))
                    nc.addNum(Double.NaN);
                else {
                    for (int i = 1; i < cutz.length; ++i) {
                        if (rite) {
                            if (x <= cutz[i]) {
                                nc.addNum(i - 1);
                                break;
                            }
                        } else if (x < cutz[i]) {
                            nc.addNum(i - 1);
                            break;
                        }
                    }
                }
            }
        }
    }.doAll(1, Vec.T_NUM, fr).outputFrame(fr.names(), domains);
    return new ValFrame(fr2);
}

Example 12 with Chunk

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

the class AstScale method apply.

@Override
public ValFrame apply(Env env, Env.StackHelp stk, AstRoot[] asts) {
    Frame fr = stk.track(asts[1].exec(env)).getFrame();
    int ncols = fr.numCols();
    // Peel out the bias/shift/mean
    double[] means;
    if (asts[2] instanceof AstNumList) {
        means = ((AstNumList) asts[2]).expand();
        if (means.length != ncols)
            throw new IllegalArgumentException("Numlist must be the same length as the columns of the Frame");
    } else {
        double d = asts[2].exec(env).getNum();
        if (// No change on means, so zero-filled
        d == 0)
            // No change on means, so zero-filled
            means = new double[ncols];
        else if (d == 1)
            means = fr.means();
        else
            throw new IllegalArgumentException("Only true or false allowed");
    }
    // Peel out the scale/stddev
    double[] mults;
    if (asts[3] instanceof AstNumList) {
        mults = ((AstNumList) asts[3]).expand();
        if (mults.length != ncols)
            throw new IllegalArgumentException("Numlist must be the same length as the columns of the Frame");
    } else {
        Val v = asts[3].exec(env);
        if (v instanceof ValFrame) {
            mults = toArray(v.getFrame().anyVec());
        } else {
            double d = v.getNum();
            if (d == 0)
                // No change on mults, so one-filled
                Arrays.fill(mults = new double[ncols], 1.0);
            else if (d == 1)
                mults = fr.mults();
            else
                throw new IllegalArgumentException("Only true or false allowed");
        }
    }
    // Update in-place.
    // Make final copy for closure
    final double[] fmeans = means;
    // Make final copy for closure
    final double[] fmults = mults;
    new MRTask() {

        @Override
        public void map(Chunk[] cs) {
            for (int i = 0; i < cs.length; i++) for (int row = 0; row < cs[i]._len; row++) cs[i].set(row, (cs[i].atd(row) - fmeans[i]) * fmults[i]);
        }
    }.doAll(fr);
    return new ValFrame(fr);
}

Example 13 with Chunk

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

the class AstSetDomain method apply.

@Override
public ValFrame apply(Env env, Env.StackHelp stk, AstRoot[] asts) {
    Frame f = stk.track(asts[1].exec(env)).getFrame();
    String[] _domains = ((AstStrList) asts[2])._strs;
    if (f.numCols() != 1)
        throw new IllegalArgumentException("Must be a single column. Got: " + f.numCols() + " columns.");
    Vec v = f.anyVec();
    if (!v.isCategorical())
        throw new IllegalArgumentException("Vector must be a factor column. Got: " + v.get_type_str());
    if (_domains != null && _domains.length != v.domain().length) {
        // in this case we want to recollect the domain and check that number of levels matches _domains
        VecUtils.CollectDomainFast t = new VecUtils.CollectDomainFast((int) v.max());
        t.doAll(v);
        final long[] dom = t.domain();
        if (dom.length != _domains.length)
            throw new IllegalArgumentException("Number of replacement factors must equal current number of levels. Current number of levels: " + dom.length + " != " + _domains.length);
        new MRTask() {

            @Override
            public void map(Chunk c) {
                for (int i = 0; i < c._len; ++i) {
                    if (!c.isNA(i)) {
                        long num = Arrays.binarySearch(dom, c.at8(i));
                        if (num < 0)
                            throw new IllegalArgumentException("Could not find the categorical value!");
                        c.set(i, num);
                    }
                }
            }
        }.doAll(v);
    }
    v.setDomain(_domains);
    DKV.put(v);
    return new ValFrame(f);
}

Example 14 with Chunk

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

the class AstBinOp method frame_op_frame.

/**
   * Auto-widen: If one frame has only 1 column, auto-widen that 1 column to
   * the rest.  Otherwise the frames must have the same column count, and
   * auto-widen element-by-element.  Short-cut if one frame has zero
   * columns.
   */
private ValFrame frame_op_frame(Frame lf, Frame rt) {
    if (lf.numRows() != rt.numRows()) {
        // special case for broadcasting a single row of data across a frame
        if (lf.numRows() == 1 || rt.numRows() == 1) {
            if (lf.numCols() != rt.numCols())
                throw new IllegalArgumentException("Frames must have same columns, found " + lf.numCols() + " columns and " + rt.numCols() + " columns.");
            return frame_op_row(lf, rt);
        } else
            throw new IllegalArgumentException("Frames must have same rows, found " + lf.numRows() + " rows and " + rt.numRows() + " rows.");
    }
    if (lf.numCols() == 0)
        return new ValFrame(lf);
    if (rt.numCols() == 0)
        return new ValFrame(rt);
    if (lf.numCols() == 1 && rt.numCols() > 1)
        return vec_op_frame(lf.vecs()[0], rt);
    if (rt.numCols() == 1 && lf.numCols() > 1)
        return frame_op_vec(lf, rt.vecs()[0]);
    if (lf.numCols() != rt.numCols())
        throw new IllegalArgumentException("Frames must have same columns, found " + lf.numCols() + " columns and " + rt.numCols() + " columns.");
    Frame res = new MRTask() {

        @Override
        public void map(Chunk[] chks, NewChunk[] cress) {
            BufferedString lfstr = new BufferedString();
            BufferedString rtstr = new BufferedString();
            assert (cress.length << 1) == chks.length;
            for (int c = 0; c < cress.length; c++) {
                Chunk clf = chks[c];
                Chunk crt = chks[c + cress.length];
                NewChunk cres = cress[c];
                if (clf.vec().isString())
                    for (int i = 0; i < clf._len; i++) cres.addNum(str_op(clf.atStr(lfstr, i), crt.atStr(rtstr, i)));
                else
                    for (int i = 0; i < clf._len; i++) cres.addNum(op(clf.atd(i), crt.atd(i)));
            }
        }
    }.doAll(lf.numCols(), Vec.T_NUM, new Frame(lf).add(rt)).outputFrame(lf._names, null);
    // Cleanup categorical misuse
    return cleanCategorical(lf, res);
}

Example 15 with Chunk

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

the class AstBinOp method frame_op_vec.

private ValFrame frame_op_vec(Frame fr, Vec vec) {
    // Already checked for same rows, non-zero frame
    Frame lf = new Frame(fr);
    lf.add("", vec);
    Frame res = new MRTask() {

        @Override
        public void map(Chunk[] chks, NewChunk[] cress) {
            assert cress.length == chks.length - 1;
            Chunk crt = chks[cress.length];
            for (int c = 0; c < cress.length; c++) {
                Chunk clf = chks[c];
                NewChunk cres = cress[c];
                for (int i = 0; i < clf._len; i++) cres.addNum(op(clf.atd(i), crt.atd(i)));
            }
        }
    }.doAll(fr.numCols(), Vec.T_NUM, lf).outputFrame(fr._names, null);
    // Cleanup categorical misuse
    return cleanCategorical(fr, res);
}