Examples with CaseScoreResult ml.shifu.shifu.container.CaseScoreResult used on opensource projects

Example 6 with CaseScoreResult

use of ml.shifu.shifu.container.CaseScoreResult in project shifu by ShifuML.

the class DataPrepareWorker method convertModelResultIntoColScore.

/*
     * Convert model result data into column-based
     * 
     * @param evalDataList
     *            evaluation result list
     * @param columnScoreListMap
     *            (column-id, List<ColumnScoreObject>)
     */
private void convertModelResultIntoColScore(List<CaseScoreResult> scoreResultList, Map<Integer, List<ColumnScoreObject>> columnScoreListMap) {
    for (CaseScoreResult scoreResult : scoreResultList) {
        Map<String, String> rawDataMap = CommonUtils.convertDataIntoMap(scoreResult.getInputData(), super.modelConfig.getDataSetDelimiter(), this.trainDataHeader);
        for (ColumnConfig config : columnConfigList) {
            if (config.isFinalSelect()) {
                ColumnScoreObject columnScore = new ColumnScoreObject(config.getColumnNum(), rawDataMap.get(config.getColumnName()));
                columnScore.setScores(scoreResult.getScores());
                columnScore.setMaxScore(scoreResult.getMaxScore());
                columnScore.setMinScore(scoreResult.getMinScore());
                columnScore.setAvgScore(scoreResult.getAvgScore());
                List<ColumnScoreObject> csList = columnScoreListMap.get(config.getColumnNum());
                if (csList == null) {
                    csList = new ArrayList<ColumnScoreObject>();
                    columnScoreListMap.put(config.getColumnNum(), csList);
                }
                csList.add(columnScore);
            }
        }
    }
}

Example 7 with CaseScoreResult

use of ml.shifu.shifu.container.CaseScoreResult in project shifu by ShifuML.

the class ModelRunner method computeNsData.

/**
 * Run model to compute score for input NS Data map
 *
 * @param rawDataNsMap
 *            - the original input, but key is wrapped by NSColumn
 * @return CaseScoreResult - model score
 */
public CaseScoreResult computeNsData(final Map<NSColumn, String> rawDataNsMap) {
    if (MapUtils.isEmpty(rawDataNsMap)) {
        return null;
    }
    CaseScoreResult scoreResult = new CaseScoreResult();
    if (this.scorer != null) {
        ScoreObject so = scorer.scoreNsData(rawDataNsMap);
        if (so == null) {
            return null;
        }
        scoreResult.setScores(so.getScores());
        scoreResult.setMaxScore(so.getMaxScore());
        scoreResult.setMinScore(so.getMinScore());
        scoreResult.setAvgScore(so.getMeanScore());
        scoreResult.setMedianScore(so.getMedianScore());
        scoreResult.setHiddenLayerScores(so.getHiddenLayerScores());
    }
    if (MapUtils.isNotEmpty(this.subScorers)) {
        if (this.isMultiThread && this.subScorers.size() > 1 && this.executorManager == null) {
            int threadPoolSize = Math.min(Runtime.getRuntime().availableProcessors(), this.subScorers.size());
            this.executorManager = new ExecutorManager<Pair<String, ScoreObject>>(threadPoolSize);
            // add a shutdown hook as a safe guard if some one not call close
            Runtime.getRuntime().addShutdownHook(new Thread(new Runnable() {

                @Override
                public void run() {
                    ModelRunner.this.executorManager.forceShutDown();
                }
            }));
            log.info("MultiThread is enabled in ModelRunner, threadPoolSize = " + threadPoolSize);
        }
        List<Callable<Pair<String, ScoreObject>>> tasks = new ArrayList<Callable<Pair<String, ScoreObject>>>(this.subScorers.size());
        Iterator<Map.Entry<String, Scorer>> iterator = this.subScorers.entrySet().iterator();
        while (iterator.hasNext()) {
            final Map.Entry<String, Scorer> entry = iterator.next();
            Callable<Pair<String, ScoreObject>> callable = new Callable<Pair<String, ScoreObject>>() {

                @Override
                public Pair<String, ScoreObject> call() {
                    String modelName = entry.getKey();
                    Scorer subScorer = entry.getValue();
                    ScoreObject so = subScorer.scoreNsData(rawDataNsMap);
                    if (so != null) {
                        return Pair.of(modelName, so);
                    } else {
                        return null;
                    }
                }
            };
            tasks.add(callable);
        }
        if (this.isMultiThread && this.subScorers.size() > 1) {
            List<Pair<String, ScoreObject>> results = this.executorManager.submitTasksAndWaitResults(tasks);
            for (Pair<String, ScoreObject> result : results) {
                if (result != null) {
                    scoreResult.addSubModelScore(result.getLeft(), result.getRight());
                }
            }
        } else {
            for (Callable<Pair<String, ScoreObject>> task : tasks) {
                Pair<String, ScoreObject> result = null;
                try {
                    result = task.call();
                } catch (Exception e) {
                // do nothing
                }
                if (result != null) {
                    scoreResult.addSubModelScore(result.getLeft(), result.getRight());
                }
            }
        }
    }
    return scoreResult;
}

Example 8 with CaseScoreResult

use of ml.shifu.shifu.container.CaseScoreResult in project shifu by ShifuML.

the class EvalNormUDF method exec.

public Tuple exec(Tuple input) throws IOException {
    if (isCsvFormat) {
        String firstCol = ((input.get(0) == null) ? "" : input.get(0).toString());
        if (this.headers[0].equals(CommonUtils.normColumnName(firstCol))) {
            // TODO what to do if the column value == column name? ...
            return null;
        }
    }
    if (this.modelRunner == null && this.isAppendScore) {
        // here to initialize modelRunner, this is moved from constructor to here to avoid OOM in client side.
        // UDF in pig client will be initialized to get some metadata issues
        @SuppressWarnings("deprecation") List<BasicML> models = ModelSpecLoaderUtils.loadBasicModels(modelConfig, evalConfig, evalConfig.getDataSet().getSource(), evalConfig.getGbtConvertToProb(), evalConfig.getGbtScoreConvertStrategy());
        this.modelRunner = new ModelRunner(modelConfig, columnConfigList, this.headers, evalConfig.getDataSet().getDataDelimiter(), models);
        this.modelRunner.setScoreScale(Integer.parseInt(this.scale));
    }
    Map<NSColumn, String> rawDataNsMap = CommonUtils.convertDataIntoNsMap(input, this.headers, this.segFilterSize);
    if (MapUtils.isEmpty(rawDataNsMap)) {
        return null;
    }
    Tuple tuple = TupleFactory.getInstance().newTuple();
    for (int i = 0; i < this.outputNames.size(); i++) {
        String name = this.outputNames.get(i);
        String raw = rawDataNsMap.get(new NSColumn(name));
        if (i == 0) {
            tuple.append(raw);
        } else if (i == 1) {
            tuple.append(StringUtils.isEmpty(raw) ? "1" : raw);
        } else if (i > 1 && i < 2 + validMetaSize) {
            // [2, 2 + validMetaSize) are meta columns
            tuple.append(raw);
        } else {
            ColumnConfig columnConfig = this.columnConfigMap.get(name);
            List<Double> normVals = Normalizer.normalize(columnConfig, raw, this.modelConfig.getNormalizeStdDevCutOff(), this.modelConfig.getNormalizeType());
            if (this.isOutputRaw) {
                tuple.append(raw);
            }
            for (Double normVal : normVals) {
                tuple.append(getOutputValue(normVal, true));
            }
        }
    }
    if (this.isAppendScore && this.modelRunner != null) {
        CaseScoreResult score = this.modelRunner.computeNsData(rawDataNsMap);
        if (this.modelRunner == null || this.modelRunner.getModelsCnt() == 0 || score == null) {
            tuple.append(-999.0);
        } else if (this.scIndex < 0) {
            tuple.append(score.getAvgScore());
        } else {
            tuple.append(score.getScores().get(this.scIndex));
        }
    }
    return tuple;
}

Example 9 with CaseScoreResult

use of ml.shifu.shifu.container.CaseScoreResult in project shifu by ShifuML.

the class FullScoreUDF method exec.

public Tuple exec(Tuple input) throws IOException {
    Map<NSColumn, String> rawDataNsMap = CommonUtils.convertDataIntoNsMap(input, this.header, 0);
    CaseScoreResult cs = modelRunner.computeNsData(rawDataNsMap);
    if (cs == null) {
        log.error("Get null result.");
        return null;
    }
    Tuple tuple = TupleFactory.getInstance().newTuple();
    tuple.append(cs.getAvgScore());
    tuple.append(cs.getMaxScore());
    tuple.append(cs.getMinScore());
    for (double score : cs.getScores()) {
        tuple.append(score);
    }
    List<String> metaList = modelConfig.getMetaColumnNames();
    for (String meta : metaList) {
        tuple.append(rawDataNsMap.get(new NSColumn(meta)));
    }
    return tuple;
}