categoricaltransformationoperator.java

一个数据挖掘软件ALPHAMINERR的整个过程的JAVA版源代码

//    			    mapping.addCategoryPair(cat, new Category(new Integer(newCat)));
//					break;
//				case CategoricalAttribute.BOOLEAN:
//				    mapping.addCategoryPair(cat, new Category(new Boolean(newCat)));
//					break;
//				case CategoricalAttribute.DATETIME_PRUDSYS:
//				    mapping.addCategoryPair(cat, new Category(new Double(newCat)));
//					break;
//				case CategoricalAttribute.DATETIME_UNIX:
//				    mapping.addCategoryPair(cat, new Category(Long.getLong(newCat)));
//				    break;
            }
        }
   
		return mapping;
	}
	
	public CategNumMapping preparecCategNumMapping(MiningDataSpecification a_MetaData, IOperatorNode a_Node) throws MiningException
	{
	    CategNumMapping mapping = new CategNumMapping();
		
		String sourceName = null;
		String dataTypeStr = null;
		int dataType = NumericAttribute.DOUBLE;
		String targetCategStr = null;
		Vector targetCateg = new Vector();		
		//get the target attribute
		sourceName = (String) a_Node.getParameterValue("target");
	 
	    //get the target data type 
	    dataTypeStr = (String) a_Node.getParameterValue("targetDataType");
	    MiningAttribute mAtt = a_MetaData.getMiningAttribute(sourceName);
	    
	    //get the target categories
	    targetCategStr = (String) a_Node.getParameterValue("targetCateg");
	  
	    dataType = Integer.valueOf(dataTypeStr).intValue();
	    targetCateg = ValueValidator.unescapeParam(targetCategStr);
	  	    
	    mapping.setSourceName(sourceName);
	    mapping.setTargetName(sourceName);
	    
        for(int i=0; i<((CategoricalAttribute)mAtt).getCategoriesNumber(); i++){
            Category cat = ((CategoricalAttribute)mAtt).getCategory(i);
            String newCat = (String)targetCateg.get(i);
            switch(dataType){
            	case NumericAttribute.DOUBLE: 
            	    if(ValueValidator.isDouble(newCat))
            	        mapping.addCategoryPair(cat, new Double(newCat));
            	    else 
            	        mapping.addCategoryPair(cat, new Double(Double.NaN));
            		break;
//            	case NumericAttribute.FLOAT:
//            	    mapping.addCategoryPair(cat, new Category(new Double(newCat)));
//        			break;
//        		case NumericAttribute.INTEGER:
//        		    mapping.addCategoryPair(cat, new Category(new Float(newCat)));
//    				break;
//    			case NumericAttribute.BOOLEAN:
//    			    mapping.addCategoryPair(cat, new Category(new Integer(newCat)));
//					break;
//				case NumericAttribute.DATETIME_PRUDSYS:
//				    mapping.addCategoryPair(cat, new Category(new Boolean(newCat)));
//					break;
//				case NumericAttribute.DATETIME_UNIX:
//				    mapping.addCategoryPair(cat, new Category(new Double(newCat)));
//					break;
            }
        }
   
		return mapping;
	}
	
	public Numerization preparecNumerization(MiningDataSpecification a_MetaData, IOperatorNode a_Node)throws MiningException
	{
	    Numerization mapping = new Numerization();
		
	    String sourceName = null;
		
		//get the target attribute
		sourceName = (String) a_Node.getParameterValue("target");
	 
	    mapping.setSourceName(sourceName);
        mapping.setTargetName(sourceName);
		return mapping;
	}
	
	
	private void validateParameters(MiningDataSpecification a_MetaData, IOperatorNode a_Node)throws AppException{
	 		
	    String value = null;
		String sourceName = null;
		int mode;
		String dataTypeStr = null;
		int dataType = CategoricalAttribute.STRING;
		String targetCategStr = null;
		Vector targetCateg = new Vector();
		MiningAttribute mAtt = null;
		boolean valid = true;
		String message = "";
		
		//get the target attribute
	    sourceName = (String) a_Node.getParameterValue("target");
	    if (sourceName == null){
	        message += Resource.srcStr("categoricalattributeMessage");
	        valid = false;
	        throw new AppException(message);
	    }
	    else {
	        mAtt = a_MetaData.getMiningAttribute(sourceName);
	        if(mAtt==null || !(mAtt instanceof CategoricalAttribute)){
	            message += Resource.srcStr("categoricalattributeMessage");
	            valid = false;
	            throw new AppException(message);
	        }
	    }
	    
	    
	    //get the mode
	    value  = (String) a_Node.getParameterValue("mode");
	    if(value ==null  || !ValueValidator.isNumeric(value)){
	        message += "Please select a target type or Auto numerization\n";
	        throw new AppException(message);
	    }
	    
	    
	    mode = Integer.valueOf(value).intValue();
	    if(mode == CATEG_MAPPING || mode == CATEG_NUM_MAPPING){
		   
		        //get the target data type
			    dataTypeStr = (String) a_Node.getParameterValue("targetDataType");
			    if (dataTypeStr==null){
			        message += Resource.srcStr("datatypeMessage");
			        valid = false;
			        throw new AppException(message);
			    }
			    
			    targetCategStr = (String) a_Node.getParameterValue("targetCateg");
			    if (targetCategStr==null){
			        message += Resource.srcStr("TargetcategoriesMessage");
			        valid = false;
			        throw new AppException(message);
			        
			    }else{
			        targetCateg = ValueValidator.unescapeParam(targetCategStr);
			        if(targetCateg.size() != ((CategoricalAttribute)mAtt).getCategoriesNumber()){
			            message += Resource.srcStr("TargetcategoriesMessage");
			            valid = false;
			            throw new AppException(message);
			        }
			    }	
			    
			    if(valid && mode == CATEG_MAPPING){
			        dataType = Integer.valueOf(dataTypeStr).intValue();
			        
			        for(int i=0; i<((CategoricalAttribute)mAtt).getCategoriesNumber(); i++){
			            //check the type
			            switch(dataType){
			            	case CategoricalAttribute.STRING:
			            	    break;
//			            	case CategoricalAttribute.DOUBLE:
//			            	    if(!ValueValidator.isDouble((String)targetCateg.get(i)))
//			            	        message += "";
//			            		break;
//			            	case CategoricalAttribute.FLOAT:
//			            	     if(!ValueValidator.isFloat((String)targetCateg.get(i)))
//			            	         message +="";
//			            		break;
//			        		case CategoricalAttribute.INTEGER:
//			        		    if(!ValueValidator.isNumeric((String)targetCateg.get(i)))
//			        		        message +="";
//		            			break;
//							case CategoricalAttribute.BOOLEAN:
//							case CategoricalAttribute.DATETIME_PRUDSYS:
//							case CategoricalAttribute.DATETIME_UNIX:
//							    break;
			            	default: 
			            	    message += Resource.srcStr("datatypeMessage")+"\n";
			            		throw new AppException(message);
			            	 
			            }
			            
			        }
			    
			    
			    }
			    
			    if(valid && mode == CATEG_NUM_MAPPING){
			        dataType = Integer.valueOf(dataTypeStr).intValue();
			        ArrayList catValues = ((CategoricalAttribute)mAtt).getValues(); 
			        
			        for(int i=0; i<((CategoricalAttribute)mAtt).getCategoriesNumber(); i++){
			            String newCat = ((String)targetCateg.get(i)).trim();
			            switch(dataType){
			            	case NumericAttribute.DOUBLE: 
			            	    if(!newCat.equals("") && !ValueValidator.isDouble(newCat)){
			            	        message += "The target for category " + ((Category)catValues.get(i)).getValue().toString() +" should be a double \n";
			            	    	throw new AppException(message);
			            	    }
			            		break;
//			            	case NumericAttribute.FLOAT:
//			            	    if(!ValueValidator.isFloat((String)targetCateg.get(i)))
//			            	         message +="";
//			            		break;
//			        		case NumericAttribute.INTEGER:
//			        		    if(!ValueValidator.isNumeric((String)targetCateg.get(i)))
//			        		        message +="";
//		            			break;
//			    			case NumericAttribute.BOOLEAN:
//			    			case NumericAttribute.DATETIME_PRUDSYS:
//			    			case NumericAttribute.DATETIME_UNIX:
//								break;
			    			default:
			    			    message += Resource.srcStr("datatypeMessage");
			    				throw new AppException(message);
			            }
			        }
			    }
		   
	    }else if (mode != NUMERIZATION){
	        message += "Please select a target type or Auto numerization\n";
	        throw new AppException(message);
	    }
		     
		
//		if(!message.equals(""))		       
//	        throw new AppException(message);
	}
	
}