dm.cs

Data Mining 算法

using System;

namespace DM.NaiveBayes
{

	/// <summary>
	/// Summary description for DM.
	/// </summary>
	public abstract class DataMining
	{

		/// <summary>
		/// Imports a training dataset from an xml file	for the Naive Bayes Classifier
		/// </summary>
		/// <param name="sampleXmlFilePath">
		/// Full path pointing to an xml file containing training data for the Naive Bayes Classifier
		/// </param>
		/// <returns>
		/// A System.Data.DataSet containing training data for the Naive Bayes Classifier
		/// </returns>
		public abstract System.Data.DataSet Import(string sampleXmlFilePath);
		
		/// <summary>
		///	 Create a statistics table 
		/// </summary>
		/// <param name="attributesXmlFilePath">
		/// Full path pointing to an xml file that describes the attributes of the training data
		/// </param>
		/// <returns>
		/// A System.Data.Dataset containing the attributes of the DataSet to train and classify
		/// </returns>
		public abstract System.Data.DataSet CreateStatistics(string attributesXmlFilePath);

		public abstract System.Data.DataSet Train(System.Data.DataSet sampleDataSet, System.Data.DataSet statisicsDataSet);

		public abstract System.Data.DataSet Classify (System.Data.DataSet newDataSet, System.Data.DataSet trainingDataSet);

		public abstract void Export(System.Data.DataSet data, string exportXmlFilePath);

		public bool Process(string sampleXmlFilePath, string attributesXmlFilePath, System.Data.DataSet newDataSet)
		{
			try
			{
				System.Data.DataSet importData = Import(sampleXmlFilePath);

				System.Data.DataSet statsData = CreateStatistics(attributesXmlFilePath);

				System.Data.DataSet trainData = Train(importData, statsData);

				System.Data.DataSet classifiedData = Classify(newDataSet, trainData);

				return true;
			}
			catch(Exception ex)
			{
				System.Diagnostics.Debug.WriteLine(ex.ToString());

				return false;
			}
		}

	}
	
	public class NaiveBayes : DataMining
	{

		private System.Data.DataSet _statistics = null;

		private System.Data.DataRow [] _classValues = null;

		private System.Data.DataRow [] _attributeValues = null;


        public override System.Data.DataSet Import(string xmlFilePath)
		{
			System.Data.DataSet importData = new System.Data.DataSet();
			
			importData.ReadXml(xmlFilePath, System.Data.XmlReadMode.InferSchema);
			 
			return importData;			
		}

		public override System.Data.DataSet CreateStatistics(string attributeValueFilePath)
		{   
			//ts.tableCollection[0].rowCollection[0][1] = "NOMINAL"  -- TYPE
			//ts.tableCollection[0].rowCollection[0][1] = "CUSTOMERINCOME"  -- NAME
			//ts.tableCollection[0].rowCollection[0][2] = "LOW"  -- VALUE

			System.Data.DataSet statisicsDataSet = new System.Data.DataSet();
			
			statisicsDataSet.ReadXml(attributeValueFilePath, System.Data.XmlReadMode.InferSchema);

			System.Data.DataRow [] cType = statisicsDataSet.Tables[0].Select("TYPE=" + "'" + "CLASS" + "'", "TYPE ASC");

			this._classValues = cType;

			this._statistics = statisicsDataSet;

			for(int count = 0; count < cType.Length; count++)
			{
				//Add Columns
				_statistics.Tables[0].Columns.Add( "D-" + cType[count][2], typeof(string) );

				_statistics.Tables[0].Columns.Add( "P-" + cType[count][2], typeof(string) );
			}

			foreach(System.Data.DataColumn dc in _statistics.Tables[0].Columns)
			{
			  System.Diagnostics.Debug.WriteLine(dc.ColumnName.ToString());
			}

			statisicsDataSet = null;

			cType = null;

			return _statistics;
		}


		public override System.Data.DataSet Train(System.Data.DataSet sampleDataSet, System.Data.DataSet statisicsDataSet)
		{
			//get the row of attributes and values 
			System.Data.DataRow [] cType = statisicsDataSet.Tables[0].Select("TYPE <> " + "'" + "CLASS" + "'", "NAME ASC");

			this._attributeValues = cType;

			System.Text.StringBuilder filters = null;

			System.Text.StringBuilder summary = null;

			string [] filter = new string [cType.Length];

			int [] total = new int [cType.Length];

			for(int count = 0; count < cType.Length; count++)
			{
				filter[count] = this._attributeValues[count][1].ToString() + " = " + "'" + this._attributeValues[count][2].ToString() + "'";

				for(int cCount = 0; cCount < this._classValues.Length; cCount++)
				{
					filters = new System.Text.StringBuilder (24);

					filters.Append(filter[count].ToString() + " and " + this._classValues[cCount][1].ToString() + " = " + "'" + this._classValues[cCount][2].ToString() + "'");

					total[count] = sampleDataSet.Tables[0].Select(filters.ToString()).Length; 

					System.Diagnostics.Debug.WriteLine("Filter : " + filters.ToString() + " Total : " + total[count].ToString());      

					summary = new System.Text.StringBuilder (24);

					summary.Append("NAME = " + "'" + this._attributeValues[count][1].ToString() + "'" + " and " + "VALUE = " + "'" + this._attributeValues[count][2].ToString() + "'");

					//update the right row in the statistics table
					System.Data.DataRow [] row = statisicsDataSet.Tables[0].Select(summary.ToString());

					System.Diagnostics.Debug.WriteLine("Summary Filter : " + summary.ToString() + " :  " + row.Length.ToString() + " rows selected" );

					//TODO: Not completed
				}
			}

			cType = null;

			filters = null;

			filter = null;
			
			return statisicsDataSet;
		}
		
		public override System.Data.DataSet Classify(System.Data.DataSet newDataSet, System.Data.DataSet trainingDataSet)
		{
			System.Diagnostics.Debug.WriteLine("TODO: Classify Method");
			
			return null;
		}

		public override void Export(System.Data.DataSet data, string exportXmlFilePath)
		{
		  System.Diagnostics.Debug.WriteLine("TODO: Export Method"); 
       	
		  return;
		}
	}

	public class Classifier : NaiveBayes
	{

	}

}