* @param output
* the output pathname String
*/
@Override
public void runJob(String input, String output, BayesParameters params) throws IOException {
Configurable client = new JobClient();
JobConf conf = new JobConf(BayesThetaNormalizerDriver.class);
conf.setJobName("Bayes Theta Normalizer Driver running over input: " + input);
conf.setOutputKeyClass(StringTuple.class);
conf.setOutputValueClass(DoubleWritable.class);
FileInputFormat.addInputPath(conf, new Path(output + "/trainer-tfIdf/trainer-tfIdf"));
Path outPath = new Path(output + "/trainer-thetaNormalizer");
FileOutputFormat.setOutputPath(conf, outPath);
// conf.setNumMapTasks(100);
// conf.setNumReduceTasks(1);
conf.setMapperClass(BayesThetaNormalizerMapper.class);
conf.setInputFormat(SequenceFileInputFormat.class);
conf.setCombinerClass(BayesThetaNormalizerReducer.class);
conf.setReducerClass(BayesThetaNormalizerReducer.class);
conf.setOutputFormat(SequenceFileOutputFormat.class);
conf.set("io.serializations", "org.apache.hadoop.io.serializer.JavaSerialization,"
+ "org.apache.hadoop.io.serializer.WritableSerialization");
// Dont ever forget this. People should keep track of how hadoop conf
// parameters and make or break a piece of code
FileSystem dfs = FileSystem.get(outPath.toUri(), conf);
if (dfs.exists(outPath)) {
dfs.delete(outPath, true);
}
Path sigmaKFiles = new Path(output + "/trainer-weights/Sigma_k/*");
Map<String,Double> labelWeightSum = SequenceFileModelReader.readLabelSums(dfs, sigmaKFiles, conf);
DefaultStringifier<Map<String,Double>> mapStringifier = new DefaultStringifier<Map<String,Double>>(conf,
GenericsUtil.getClass(labelWeightSum));
String labelWeightSumString = mapStringifier.toString(labelWeightSum);
log.info("Sigma_k for Each Label");
Map<String,Double> c = mapStringifier.fromString(labelWeightSumString);
log.info("{}", c);
conf.set("cnaivebayes.sigma_k", labelWeightSumString);
Path sigmaJSigmaKFile = new Path(output + "/trainer-weights/Sigma_kSigma_j/*");
double sigmaJSigmaK = SequenceFileModelReader.readSigmaJSigmaK(dfs, sigmaJSigmaKFile, conf);
DefaultStringifier<Double> stringifier = new DefaultStringifier<Double>(conf, Double.class);
String sigmaJSigmaKString = stringifier.toString(sigmaJSigmaK);
log.info("Sigma_kSigma_j for each Label and for each Features");
double retSigmaJSigmaK = stringifier.fromString(sigmaJSigmaKString);
log.info("{}", retSigmaJSigmaK);
conf.set("cnaivebayes.sigma_jSigma_k", sigmaJSigmaKString);
Path vocabCountFile = new Path(output + "/trainer-tfIdf/trainer-vocabCount/*");
double vocabCount = SequenceFileModelReader.readVocabCount(dfs, vocabCountFile, conf);
String vocabCountString = stringifier.toString(vocabCount);
log.info("Vocabulary Count");
conf.set("cnaivebayes.vocabCount", vocabCountString);
double retvocabCount = stringifier.fromString(vocabCountString);
log.info("{}", retvocabCount);
conf.set("bayes.parameters", params.toString());
conf.set("output.table", output);
client.setConf(conf);
JobClient.runJob(conf);
}