Examples with AstStrList water.rapids.ast.params.AstStrList used on opensource projects

Example 1 with AstStrList

use of water.rapids.ast.params.AstStrList 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 2 with AstStrList

use of water.rapids.ast.params.AstStrList in project h2o-3 by h2oai.

the class AstMatch method apply.

@Override
public ValFrame apply(Env env, Env.StackHelp stk, AstRoot[] asts) {
    Frame fr = stk.track(asts[1].exec(env)).getFrame();
    if ((fr.numCols() != 1) || !(fr.anyVec().isCategorical() || fr.anyVec().isString()))
        throw new IllegalArgumentException("can only match on a single categorical/string column.");
    final MRTask<?> matchTask;
    double noMatch = asts[3].exec(env).getNum();
    if (asts[2] instanceof AstNumList) {
        matchTask = new NumMatchTask(((AstNumList) asts[2]).sort().expand(), noMatch);
    } else if (asts[2] instanceof AstNum) {
        matchTask = new NumMatchTask(new double[] { asts[2].exec(env).getNum() }, noMatch);
    } else if (asts[2] instanceof AstStrList) {
        String[] values = ((AstStrList) asts[2])._strs;
        Arrays.sort(values);
        matchTask = fr.anyVec().isString() ? new StrMatchTask(values, noMatch) : new CatMatchTask(values, noMatch);
    } else if (asts[2] instanceof AstStr) {
        String[] values = new String[] { asts[2].exec(env).getStr() };
        matchTask = fr.anyVec().isString() ? new StrMatchTask(values, noMatch) : new CatMatchTask(values, noMatch);
    } else
        throw new IllegalArgumentException("Expected numbers/strings. Got: " + asts[2].getClass());
    Frame result = matchTask.doAll(Vec.T_NUM, fr.anyVec()).outputFrame();
    return new ValFrame(result);
}

Example 3 with AstStrList

use of water.rapids.ast.params.AstStrList in project h2o-3 by h2oai.

the class AstCountMatches method apply.

@Override
public ValFrame apply(Env env, Env.StackHelp stk, AstRoot[] asts) {
    Frame fr = stk.track(asts[1].exec(env)).getFrame();
    final String[] pattern = asts[2] instanceof AstStrList ? ((AstStrList) asts[2])._strs : new String[] { asts[2].exec(env).getStr() };
    // Type check
    for (Vec v : fr.vecs()) if (!(v.isCategorical() || v.isString()))
        throw new IllegalArgumentException("countmatches() requires a string or categorical column. " + "Received " + fr.anyVec().get_type_str() + ". Please convert column to a string or categorical first.");
    // Transform each vec
    Vec[] nvs = new Vec[fr.numCols()];
    int i = 0;
    for (Vec v : fr.vecs()) {
        if (v.isCategorical())
            nvs[i] = countMatchesCategoricalCol(v, pattern);
        else
            nvs[i] = countMatchesStringCol(v, pattern);
        i++;
    }
    return new ValFrame(new Frame(nvs));
}

Example 4 with AstStrList

use of water.rapids.ast.params.AstStrList in project h2o-3 by h2oai.

the class AstImpute method apply.

// (h2o.impute data col method combine_method groupby groupByFrame values)
@Override
public Val apply(Env env, Env.StackHelp stk, AstRoot[] asts) {
    // Argument parsing and sanity checking
    // Whole frame being imputed
    Frame fr = stk.track(asts[1].exec(env)).getFrame();
    // Column within frame being imputed
    final int col = (int) asts[2].exec(env).getNum();
    if (col >= fr.numCols())
        throw new IllegalArgumentException("Column not -1 or in range 0 to " + fr.numCols());
    final boolean doAllVecs = col == -1;
    final Vec vec = doAllVecs ? null : fr.vec(col);
    // Technique used for imputation
    AstRoot method = null;
    boolean ffill0 = false, bfill0 = false;
    switch(asts[3].exec(env).getStr().toUpperCase()) {
        case "MEAN":
            method = new AstMean();
            break;
        case "MEDIAN":
            method = new AstMedian();
            break;
        case "MODE":
            method = new AstMode();
            break;
        case "FFILL":
            ffill0 = true;
            break;
        case "BFILL":
            bfill0 = true;
            break;
        default:
            throw new IllegalArgumentException("Method must be one of mean, median or mode");
    }
    // Only for median, how is the median computed on even sample sizes?
    QuantileModel.CombineMethod combine = QuantileModel.CombineMethod.valueOf(asts[4].exec(env).getStr().toUpperCase());
    // Group-by columns.  Empty is allowed, and perfectly normal.
    AstRoot ast = asts[5];
    AstNumList by2;
    if (ast instanceof AstNumList)
        by2 = (AstNumList) ast;
    else if (ast instanceof AstNum)
        by2 = new AstNumList(((AstNum) ast).getNum());
    else if (ast instanceof AstStrList) {
        String[] names = ((AstStrList) ast)._strs;
        double[] list = new double[names.length];
        int i = 0;
        for (String name : ((AstStrList) ast)._strs) list[i++] = fr.find(name);
        Arrays.sort(list);
        by2 = new AstNumList(list);
    } else
        throw new IllegalArgumentException("Requires a number-list, but found a " + ast.getClass());
    Frame groupByFrame = asts[6].str().equals("_") ? null : stk.track(asts[6].exec(env)).getFrame();
    AstRoot vals = asts[7];
    AstNumList values;
    if (vals instanceof AstNumList)
        values = (AstNumList) vals;
    else if (vals instanceof AstNum)
        values = new AstNumList(((AstNum) vals).getNum());
    else
        values = null;
    boolean doGrpBy = !by2.isEmpty() || groupByFrame != null;
    // Compute the imputed value per-group.  Empty groups are allowed and OK.
    IcedHashMap<AstGroup.G, Freezable[]> group_impute_map;
    if (!doGrpBy) {
        // Skip the grouping work
        if (ffill0 || bfill0) {
            // do a forward/backward fill on the NA
            // TODO: requires chk.previousNonNA and chk.nextNonNA style methods (which may go across chk boundaries)s
            final boolean ffill = ffill0;
            final boolean bfill = bfill0;
            throw H2O.unimpl("No ffill or bfill imputation supported");
        //        new MRTask() {
        //          @Override public void map(Chunk[] cs) {
        //            int len=cs[0]._len; // end of this chk
        //            long start=cs[0].start();  // absolute beginning of chk s.t. start-1 bleeds into previous chk
        //            long absEnd = start+len;   // absolute end of the chk s.t. absEnd+1 bleeds into next chk
        //            for(int c=0;c<cs.length;++c )
        //              for(int r=0;r<cs[0]._len;++r ) {
        //                if( cs[c].isNA(r) ) {
        //                  if( r > 0 && r < len-1 ) {
        //                    cs[c].set(r,ffill?)
        //                  }
        //                }
        //              }
        //          }
        //        }.doAll(doAllVecs?fr:new Frame(vec));
        //        return new ValNum(Double.NaN);
        } else {
            final double[] res = values == null ? new double[fr.numCols()] : values.expand();
            if (values == null) {
                // fill up res if no values supplied user, common case
                if (doAllVecs) {
                    for (int i = 0; i < res.length; ++i) if (fr.vec(i).isNumeric() || fr.vec(i).isCategorical())
                        res[i] = fr.vec(i).isNumeric() ? fr.vec(i).mean() : ArrayUtils.maxIndex(fr.vec(i).bins());
                } else {
                    Arrays.fill(res, Double.NaN);
                    if (method instanceof AstMean)
                        res[col] = vec.mean();
                    if (method instanceof AstMedian)
                        res[col] = AstMedian.median(new Frame(vec), combine);
                    if (method instanceof AstMode)
                        res[col] = AstMode.mode(vec);
                }
            }
            new MRTask() {

                @Override
                public void map(Chunk[] cs) {
                    int len = cs[0]._len;
                    // run down each chk
                    for (int c = 0; c < cs.length; ++c) if (!Double.isNaN(res[c]))
                        for (int row = 0; row < len; ++row) if (cs[c].isNA(row))
                            cs[c].set(row, res[c]);
                }
            }.doAll(fr);
            return new ValNums(res);
        }
    } else {
        if (col >= fr.numCols())
            throw new IllegalArgumentException("Column not -1 or in range 0 to " + fr.numCols());
        Frame imputes = groupByFrame;
        if (imputes == null) {
            // Build and run a GroupBy command
            AstGroup ast_grp = new AstGroup();
            // simple case where user specified a column... col == -1 means do all columns
            if (doAllVecs) {
                AstRoot[] aggs = new AstRoot[(int) (3 + 3 * (fr.numCols() - by2.cnt()))];
                aggs[0] = ast_grp;
                aggs[1] = new AstFrame(fr);
                aggs[2] = by2;
                int c = 3;
                for (int i = 0; i < fr.numCols(); ++i) {
                    if (!by2.has(i) && (fr.vec(i).isCategorical() || fr.vec(i).isNumeric())) {
                        aggs[c] = fr.vec(i).isNumeric() ? new AstMean() : new AstMode();
                        aggs[c + 1] = new AstNumList(i, i + 1);
                        aggs[c + 2] = new AstStr("rm");
                        c += 3;
                    }
                }
                imputes = ast_grp.apply(env, stk, aggs).getFrame();
            } else
                imputes = ast_grp.apply(env, stk, new AstRoot[] { ast_grp, new AstFrame(fr), by2, /**/
                method, new AstNumList(col, col + 1), new AstStr("rm") }).getFrame();
        }
        if (// >2 makes it ambiguous which columns are groupby cols and which are aggs, throw IAE
        by2.isEmpty() && imputes.numCols() > 2)
            throw new IllegalArgumentException("Ambiguous group-by frame. Supply the `by` columns to proceed.");
        final int[] bycols0 = ArrayUtils.seq(0, Math.max((int) by2.cnt(), 1));
        group_impute_map = new Gather(by2.expand4(), bycols0, fr.numCols(), col).doAll(imputes)._group_impute_map;
        // Now walk over the data, replace NAs with the imputed results
        final IcedHashMap<AstGroup.G, Freezable[]> final_group_impute_map = group_impute_map;
        if (by2.isEmpty()) {
            int[] byCols = new int[imputes.numCols() - 1];
            for (int i = 0; i < byCols.length; ++i) byCols[i] = fr.find(imputes.name(i));
            by2 = new AstNumList(byCols);
        }
        final int[] bycols = by2.expand4();
        new MRTask() {

            @Override
            public void map(Chunk[] cs) {
                Set<Integer> _bycolz = new HashSet<>();
                for (int b : bycols) _bycolz.add(b);
                AstGroup.G g = new AstGroup.G(bycols.length, null);
                for (int row = 0; row < cs[0]._len; row++) for (int c = 0; c < cs.length; ++c) if (!_bycolz.contains(c))
                    if (cs[c].isNA(row))
                        cs[c].set(row, ((IcedDouble) final_group_impute_map.get(g.fill(row, cs, bycols))[c])._val);
            }
        }.doAll(fr);
        return new ValFrame(imputes);
    }
}

Example 5 with AstStrList

use of water.rapids.ast.params.AstStrList in project h2o-3 by h2oai.

the class AstColNames method apply.

@Override
public ValFrame apply(Env env, Env.StackHelp stk, AstRoot[] asts) {
    Frame fr = stk.track(asts[1].exec(env)).getFrame();
    if (asts[2] instanceof AstNumList) {
        if (!(asts[3] instanceof AstStrList))
            throw new IllegalArgumentException("Column naming requires a string-list, but found a " + asts[3].getClass());
        AstNumList cols = ((AstNumList) asts[2]);
        AstStrList nams = ((AstStrList) asts[3]);
        int[] d = cols.expand4();
        if (d.length != nams._strs.length)
            throw new IllegalArgumentException("Must have the same number of column choices as names");
        for (int i = 0; i < d.length; i++) fr._names[d[i]] = nams._strs[i];
    } else if ((asts[2] instanceof AstNum)) {
        int col = (int) (asts[2].exec(env).getNum());
        String name = asts[3].exec(env).getStr();
        fr._names[col] = name;
    } else
        throw new IllegalArgumentException("Column naming requires a number-list, but found a " + asts[2].getClass());
    // Update names in DKV
    if (fr._key != null)
        DKV.put(fr);
    return new ValFrame(fr);
}