tools.cc

一种聚类算法,名字是cocluster

    case MSSRCC_II_ALGORITHM:
      myCLA.algorithmType = MINIMUM_SUM_SQUARE_RESIDUE_II_CC;
      break;
    default:
      cout << "  !!! Invalid algorithm type: " << *argv << " !!!" << endl << endl;
      validCLA = false;
      break;
  }
  if (!validCLA){
    printAlgorithmType();
    myCLA.numInvalidCLA++;
  }
  return argv;
}


// get # of column clusters from command-line...
char **getColClusterNum(int argc, char **argv, commandLineArgument &myCLA)
{
  bool validCLA = true;
  if (*(argv+1) == NULL){
    cout << "  !!! Invalid # of column cluster(s): (# of column cluster(s) should be specified.) !!!" << endl << endl;
    validCLA = false;
  }
  if (validCLA && !isdigit((*(argv+1))[0])){
    cout << "  !!! Invalid # of column cluster(s): " << *(argv+1) << " !!!" << endl << endl;
    validCLA = false;
  }
  int tempNum = atoi(*(argv+1));
  if (tempNum <= 0){
    cout << "  !!! Invalid # of column cluster(s): (# of column cluster(s) should be a positive integer.): " << tempNum << " !!!" << endl;
    validCLA = false;
  }
  if (validCLA)
    myCLA.numColCluster = atoi(*(++argv));
  if (!validCLA){
    printColClusterNum();
    myCLA.numInvalidCLA++;
  }
  return argv;
}


// get dump-level from command-line...
char **getDumpLevel(int argc, char **argv, commandLineArgument &myCLA)
{
  bool validCLA = true;
  if (*(argv+1) == NULL){
    cout << "  !!! Invalid dump level: (dump level should be specified.) !!!" << endl << endl;
    validCLA = false;
  }
  if (validCLA && !isdigit((*(argv+1))[0])){
    cout << "  !!! Invalid dump level: " << *(argv+1) << " !!!" << endl << endl;
    validCLA = false;
  }
  if (validCLA){
    myCLA.dumpLevel = atoi(*(++argv));
    if ((myCLA.dumpLevel < MINIMUM_DUMP_LEVEL) || (myCLA.dumpLevel > MAXIMUM_DUMP_LEVEL)){
      cout << "  !!! Invalid dump level: " << myCLA.dumpLevel << " !!!" << endl << endl;
      validCLA = false;
    }
  }
  if (validCLA && (myCLA.dumpLevel == MAXIMUM_DUMP_LEVEL)){
    switch (toupper((*(++argv))[0])){
      case APPEND_OUT:
        myCLA.dumpAccessMode = APPEND_MODE;
	break;
      case OUTPUT_OUT:
        myCLA.dumpAccessMode = OUTPUT_MODE;
	break;
      default:
	cout << "  !!! Invalid dump file access mode: " << *argv << " !!!" << endl << endl;
        validCLA = false;
        break;
    }
    if (validCLA && (*(argv+1) == NULL)){
      cout << "  !!! Invalid dump filename: (dump filename should be specified.) !!!" << endl << endl;
      validCLA = false;
    }
    if (validCLA && (isdigit((*(argv+1))[0]) || ((*(argv+1))[0] == '-'))){
      cout << "  Invalid dump filename: " << *(argv+1) << endl << endl;
      validCLA = false;
    }
    if (validCLA)
      strcpy(myCLA.dumpFilename, *(++argv));
  }
  if (!validCLA){
    printOutputLabelType();
    myCLA.numInvalidCLA++;
  }
  return argv;
}


// get information of showing a CM and each cluster's statistics...
char **getShowingEachCluster(int argc, char **argv, commandLineArgument &myCLA)
{
  bool validCLA = true;
  if (*(argv+1) == NULL){
    cout << "  !!! Invalid parameter: (0 or 1 should be specified.) !!!" << endl << endl;
    validCLA = false;
  }
  if (validCLA && !isdigit((*(argv+1))[0])){
    cout << "  !!! Invalid parameter: " << *(argv+1) << " !!!" << endl << endl;
    validCLA = false;
  }
  if (validCLA){
    switch (atoi(*(++argv))){
      case ON:
        myCLA.showingEachCluster = true;
	break;
      case OFF:
        myCLA.showingEachCluster = false;
	break;
      default:
        cout << " !!! Invalid parameter: (0 or 1 should be specified.): " << myCLA.numRun << " !!!" << endl << endl;
        validCLA = false;
	break;
    }
  }
  if (!validCLA){
    printShowingEachCluster();
    myCLA.numInvalidCLA++;
  }
  return argv;
}


// get input file information from command-line...
char **getInputFileInformation(int argc, char **argv, commandLineArgument &myCLA)
{
  bool validCLA = true;
  switch (toupper((*(++argv))[0])){
    case DENSE_INPUT:
      myCLA.inputMatrixType = DENSE_MATRIX;
      switch (toupper((*(++argv))[0])){
	case SEPARATE_INPUT:
	  myCLA.inputFormatType = DIM_MATRIX_SEPARATE_FORMAT;
	  break;
        case TOGETHER_INPUT:
          myCLA.inputFormatType = DIM_MATRIX_TOGETHER_FORMAT;
	  break;
        default:
          cout << "  !!! Invalid dense matrix format type: " << *argv << " !!!" << endl << endl;
          validCLA = false;
	  break;
      }
      break;
    case SPARSE_INPUT:
      myCLA.inputMatrixType = SPARSE_MATRIX;
      switch (toupper((*(++argv))[0])){
	case TFN_INPUT:
	  strcpy(myCLA.scalingType, TFN_SCALING);	// In fact, it's not used.
	  myCLA.inputFormatType = TFN_FILE_FORMAT;
	  break;
        case TXX_INPUT:
          strcpy(myCLA.scalingType, TXX_SCALING);	// In fact, it's not used
	  myCLA.inputFormatType = TXX_FILE_FORMAT;
	  break;
        default:
          cout << "  !!! Invalid sparse matrix format type: " << *argv << " !!!" << endl << endl;
          validCLA = false;
	  break;
      }
      break;
    default:  
      cout << "  !!! Invalid input matrix argument(s): " << *argv << " !!!" << endl << endl;
      validCLA = false;
      break;
  }
  if (validCLA && ((*(++argv))[0] != '-')){
    sprintf(myCLA.inputFilename, "%s",*argv);
    if (strcmp(myCLA.coclusterFilename, "") == 0)
      extractFilename(*argv, myCLA.coclusterFilename);
    strcpy(myCLA.dumpFilename, myCLA.coclusterFilename);
    strcpy(myCLA.objectiveFilename, myCLA.coclusterFilename);
    myCLA.havingArgument = true;
  } else {
     cout << "  !!! Invalid input matrix argument(s): " << *argv << " !!!" << endl << endl;
     validCLA = false;
  }
  if (!validCLA){
    printInputMatrixType();
    myCLA.numInvalidCLA++;
  }
  return argv;
}


// get information of computing one-way clustering objective function values from command-line...
char **getComputingOneWayObjective(int argc, char **argv, commandLineArgument &myCLA)
{
  bool validCLA = true;
  if (*(argv+1) == NULL){
    cout << "  !!! Invalid parameter: (0 or 1 should be specified.) !!!" << endl << endl;
    validCLA = false;
  }
  if (validCLA && !isdigit((*(argv+1))[0])){
    cout << "  !!! Invalid parameter: " << *(argv+1) << " !!!" << endl << endl;
    validCLA = false;
  }
  if (validCLA){
    switch (atoi(*(++argv))){
      case ON:
        myCLA.computingOneWayObjective = true;
	break;
      case OFF:
        myCLA.computingOneWayObjective = false;
	break;
      default:
        cout << " !!! Invalid parameter: (0 or 1 should be specified.): " << myCLA.numRun << " !!!" << endl << endl;
        validCLA = false;
	break;
    }
  }
  if (!validCLA){
    printComputingOneWayObject();
    myCLA.numInvalidCLA++;
  }
  return argv;
}


// get class (label) file information from command-line...	  
char **getClassFileInformation(int argc, char **argv, commandLineArgument &myCLA)
{
  bool validCLA = true;
  switch (toupper((*(++argv))[0])){
    case BOTH_LABEL:
      myCLA.externalValidityType = BOTH_EXTERNAL_VALIDITY;
      break;
    case COL_LABEL:
      myCLA.externalValidityType = COL_EXTERNAL_VALIDITY;
      break;
    case ROW_LABEL:
      myCLA.externalValidityType = ROW_EXTERNAL_VALIDITY;
      break;
    default:
      cout << "  !!! Invalid class matrix selection: " << *argv << " !!!" << endl << endl;
      validCLA = false;
      break;
  }
  if (validCLA){
    switch ((*(++argv))[0]){
      case LABEL_FROM_0:
        myCLA.classOffsetType = START_FROM_0;
        break;
      case LABEL_FROM_1:
        myCLA.classOffsetType = START_FROM_1;
        break;
      default:
        cout << "  !!! Invalid class matrix offset type: " << *argv << " !!!" << endl << endl;
        validCLA = false;
        break;
    }
  }
  if (validCLA && (*(argv+1) == NULL)){
    cout << "  !!! Invalid class filename: (class filename should be specified.) !!!" << endl << endl;
    validCLA = false;
  }
  if (validCLA && (isdigit((*(argv+1))[0]) || ((*(argv+1))[0] == '-'))){
    cout << "  !!! Invalid class filename: " << *(argv+1) << " !!!" << endl << endl;
    validCLA = false;
  }
  if (validCLA){
    switch (myCLA.externalValidityType){
      case BOTH_EXTERNAL_VALIDITY:
        strcpy(myCLA.bothClassFilename, *(++argv));
        break;
      case COL_EXTERNAL_VALIDITY:
        strcpy(myCLA.colClassFilename, *(++argv));
        break;
      case ROW_EXTERNAL_VALIDITY:
        strcpy(myCLA.rowClassFilename, *(++argv));
        break;
      default:
        validCLA = false;
	break;
    }
  }      
  if (!validCLA){
    printClassLabelSelection();
    myCLA.numInvalidCLA++;
  }
  return argv;
}


/*
// get length of local search from command-line...
char **getLocalSearchLength(int argc, char **argv, commandLineArgument &myCLA)
{
  myCLA.localSearchLength = atoi(*(++argv));
  return argv;
}
*/

// get type of smoothing from command-line...
char **getSmoothingType(int argc, char **argv, commandLineArgument &myCLA)
{
  bool validCLA = true;
  char tempSmoothingType = toupper((*(++argv))[0]);
  switch (tempSmoothingType){
    case ANNEALING_TYPE:
      break;
    case MAXIMUM_ENTROPY_TYPE:
      myCLA.smoothingType = MAXIMUM_ENTROPY_SMOOTHING;
      break;
    case NO_TYPE:
      myCLA.smoothingType = NO_SMOOTHING;
      break;
    case UNIFORM_TYPE:
      myCLA.smoothingType = UNIFORM_SMOOTHING;
      break;
    default:
      cout << "  !!! Invalid smoothing type: " << *argv << " !!!" << endl << endl;
      validCLA = false;
      break;
  }
  if (validCLA && (tempSmoothingType != NO_TYPE)){
    if (*(argv+1) == NULL){
      cout << "  !!! Invalid row smoothing magnitude: (row smoothing magnitude should be specified.) !!!" << endl << endl;
      validCLA = false;
    }
    if (validCLA && (!isdigit((*(argv+1))[0]) && ((*(argv+1))[0] != '.'))){
      cout << "  !!! Invalid row smoothing magnitude: " << *(argv+1) << " !!!" << endl << endl;
      validCLA = false;
    }
    if (validCLA && (*(argv+2) == NULL)){
      cout << "  !!! Invalid col smoothing magnitude: (col smoothing magnitude should be specified.) !!!" << endl << endl;
      validCLA = false;
    }
    if (validCLA && (!isdigit((*(argv+2))[0]) && ((*(argv+2))[0] != '.'))){
      cout << "  !!! Invalid col smoothing magnitude: " << *(argv+2) << " !!!" << endl << endl;
      validCLA = false;
    }
    if (validCLA){
      switch (tempSmoothingType){
        case ANNEALING_TYPE:
          myCLA.rowAnnealingFactor = atof(*(++argv));
          myCLA.colAnnealingFactor = atof(*(++argv));
	  break;
        case MAXIMUM_ENTROPY_TYPE:
        case UNIFORM_TYPE:
          myCLA.rowSmoothingFactor = atof(*(++argv));
          myCLA.colSmoothingFactor = atof(*(++argv));
          break;
        case NO_TYPE:
          break;
        default:
          cout << "  !!! Invalid smoothing parameter(s): " << *argv << " !!!" << endl << endl;
          validCLA = false;
        break;
      }
    }
  }  
  if (!validCLA){
    printSmoothingType();
    myCLA.numInvalidCLA++;
  }
  return argv;
}


// get number of co-clustering running(s) from command-line...
char **getRunNum(int argc, char **argv, commandLineArgument &myCLA)
{
  bool validCLA = true;
  if (*(argv+1) == NULL){
    cout << "  !!! Invalid # of run(s): (# of runs should be specified.) !!!" << endl << endl;
    validCLA = false;
  }
  if (validCLA && !isdigit((*(argv+1))[0])){