Examples with MRTask water.MRTask used on opensource projects

Example 6 with MRTask

use of water.MRTask in project h2o-3 by h2oai.

the class AstGroup 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();
    AstNumList groupby = check(ncols, asts[2]);
    final int[] gbCols = groupby.expand4();
    // Count of aggregates; knock off the first 4 ASTs (GB data [group-by] [order-by]...),
    // then count by triples.
    int naggs = (asts.length - 3) / 3;
    final AGG[] aggs = new AGG[naggs];
    for (int idx = 3; idx < asts.length; idx += 3) {
        Val v = asts[idx].exec(env);
        String fn = v instanceof ValFun ? v.getFun().str() : v.getStr();
        FCN fcn = FCN.valueOf(fn);
        AstNumList col = check(ncols, asts[idx + 1]);
        if (col.cnt() != 1)
            throw new IllegalArgumentException("Group-By functions take only a single column");
        // Aggregate column
        int agg_col = (int) col.min();
        if (fcn == FCN.mode && !fr.vec(agg_col).isCategorical())
            throw new IllegalArgumentException("Mode only allowed on categorical columns");
        NAHandling na = NAHandling.valueOf(asts[idx + 2].exec(env).getStr().toUpperCase());
        aggs[(idx - 3) / 3] = new AGG(fcn, agg_col, na, (int) fr.vec(agg_col).max() + 1);
    }
    // do the group by work now
    IcedHashMap<G, String> gss = doGroups(fr, gbCols, aggs);
    final G[] grps = gss.keySet().toArray(new G[gss.size()]);
    // apply an ORDER by here...
    if (gbCols.length > 0)
        Arrays.sort(grps, new java.util.Comparator<G>() {

            // Compare 2 groups.  Iterate down _gs, stop when _gs[i] > that._gs[i],
            // or _gs[i] < that._gs[i].  Order by various columns specified by
            // gbCols.  NaN is treated as least
            @Override
            public int compare(G g1, G g2) {
                for (int i = 0; i < gbCols.length; i++) {
                    if (Double.isNaN(g1._gs[i]) && !Double.isNaN(g2._gs[i]))
                        return -1;
                    if (!Double.isNaN(g1._gs[i]) && Double.isNaN(g2._gs[i]))
                        return 1;
                    if (g1._gs[i] != g2._gs[i])
                        return g1._gs[i] < g2._gs[i] ? -1 : 1;
                }
                return 0;
            }

            // I do not believe sort() calls equals() at this time, so no need to implement
            @Override
            public boolean equals(Object o) {
                throw H2O.unimpl();
            }
        });
    // Build the output!
    String[] fcnames = new String[aggs.length];
    for (int i = 0; i < aggs.length; i++) fcnames[i] = aggs[i]._fcn.toString() + "_" + fr.name(aggs[i]._col);
    MRTask mrfill = new MRTask() {

        @Override
        public void map(Chunk[] c, NewChunk[] ncs) {
            int start = (int) c[0].start();
            for (int i = 0; i < c[0]._len; ++i) {
                // One Group per row
                G g = grps[i + start];
                int j;
                for (// The Group Key, as a row
                j = 0; // The Group Key, as a row
                j < g._gs.length; // The Group Key, as a row
                j++) ncs[j].addNum(g._gs[j]);
                for (int a = 0; a < aggs.length; a++) ncs[j++].addNum(aggs[a]._fcn.postPass(g._dss[a], g._ns[a]));
            }
        }
    };
    Frame f = buildOutput(gbCols, naggs, fr, fcnames, grps.length, mrfill);
    return new ValFrame(f);
}

Example 7 with MRTask

use of water.MRTask 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 8 with MRTask

use of water.MRTask 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 9 with MRTask

use of water.MRTask 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 10 with MRTask

use of water.MRTask 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);
}