Imports System.IO
Public Class Form1
Private Sub Button1_Click(ByVal sender As System.Object, ByVal e As System.EventArgs) Handles Button1.Click
TextBox3.Text = ""
TextBox2.Text = ""
'属性个数
Dim xmcount As Integer = 0
Dim mysr As StreamReader
Dim strline As String
'是否正例
Dim iszl As Boolean
Dim mystr As String()
'标题名称
Dim xmname() As String
'变型空间
Dim bxkj As New ArrayList()
Dim tempsuppose As String()
'一般假设
Dim common_suppose As New ArrayList()
'特殊假设
Dim specific_suppose As String()
'读取样本文件
mysr = New StreamReader("e:\mydata.txt", System.Text.Encoding.Default)
'第一行是标题
strline = mysr.ReadLine()
TextBox1.Clear()
mystr = strline.Split(New Char() {",", "，"})
'取属性个数
xmcount = mystr.GetUpperBound(0)
ReDim xmname(xmcount)
'初始化假设数组
ReDim specific_suppose(xmcount)
ReDim tempsuppose(xmcount)
For my_i As Integer = 0 To xmcount
specific_suppose(my_i) = Nothing
Next
For my_i As Integer = 0 To xmcount
tempsuppose(my_i) = "?"
Next
common_suppose.Add(tempsuppose)
'取得标题内容
xmname = mystr
Dim i As Integer = 0
For Each s As String In xmname
If s.Trim() <> "" Then
TextBox1.AppendText(s & Space(5))
End If
Next s
'分析样本每一行
While (Not mysr.EndOfStream)
strline = mysr.ReadLine()
If Trim(strline) = "" Then Exit While
TextBox1.AppendText(vbCrLf)
mystr = strline.Split(New Char() {",", "，"})
'取样本性质
If Trim(mystr(xmcount + 1)) = "是" Then
iszl = True
Else
iszl = False
End If
TextBox1.AppendText(mystr(0) & Space(5))
'读取一行中的每个属性,最后一个属性指示了样本是正例还是反例，不能做为分析内容
For my_i As Integer = 1 To xmcount
TextBox1.AppendText(mystr(my_i) & Space(5))
If iszl Then
'正例
'设置特殊假设
If IsNothing(specific_suppose(my_i)) Then
'如果特殊假设的该属性未设定，则设定为样本的该属性值
specific_suppose(my_i) = mystr(my_i)
ElseIf specific_suppose(my_i) <> "?" And specific_suppose(my_i) <> mystr(my_i) Then
'如果特殊假设的该属性已设定，且不为当前样本值，则设定为？，可以接受任何值
specific_suppose(my_i) = "?"
End If
'设置一般假设
For my_ii As Integer = 0 To common_suppose.Count - 1
tempsuppose = TryCast(common_suppose.Item(my_ii), String())
If Not IsNothing(tempsuppose) Then
If tempsuppose(my_i) <> "?" And tempsuppose(my_i) <> mystr(my_i) Then
common_suppose.Item(my_ii) = Nothing
End If
End If
Next
Else
'反例
'设置一般假设
For my_ii As Integer = 0 To common_suppose.Count - 1
tempsuppose = TryCast(common_suppose.Item(my_ii), String())
If Not IsNothing(tempsuppose) Then
If mystr(my_i) = tempsuppose(my_i) Then
common_suppose.Item(my_ii) = Nothing
End If
End If
Next
tempsuppose = Nothing
ReDim tempsuppose(xmcount)
For my_iii As Integer = 1 To xmcount
If my_iii <> my_i Then
tempsuppose(my_iii) = "?"
Else
tempsuppose(my_iii) = specific_suppose(my_i)
End If
Next
'如果没有重复元素则增加
Dim iscf As Boolean
For my_ii As Integer = 0 To common_suppose.Count - 1
iscf = True
Dim mytempsuppose As String() = TryCast(common_suppose.Item(my_ii), String())
If Not IsNothing(mytempsuppose) Then
For mmm_i As Integer = 1 To xmcount
If Trim(tempsuppose(mmm_i)) <> Trim(mytempsuppose(mmm_i)) Then
iscf = False
End If
Next
End If
If iscf And (Not IsNothing(mytempsuppose)) Then
Exit For
End If
Next
If Not iscf Then
common_suppose.Add(tempsuppose)
End If
End If
Next my_i
TextBox1.AppendText(mystr(xmcount + 1) & Space(6))
If Not iszl Then
Dim isdel As Boolean
For my_ii As Integer = 0 To common_suppose.Count - 1
isdel = True
tempsuppose = TryCast(common_suppose.Item(my_ii), String())
If Not IsNothing(tempsuppose) Then
For mm_i As Integer = 1 To xmcount
If Trim(tempsuppose(mm_i)) <> "?" Then
isdel = False
End If
Next
End If
If isdel Then
common_suppose.Item(my_ii) = Nothing
End If
Next
End If
End While
For my_i As Integer = 1 To xmcount
TextBox2.Text &= specific_suppose(my_i) & Space(6)
Next
For my_iiii As Integer = 0 To common_suppose.Count - 1
If Not IsNothing(common_suppose.Item(my_iiii)) Then
tempsuppose = TryCast(common_suppose.Item(my_iiii), String())
For m_i As Integer = 1 To xmcount
TextBox3.Text &= tempsuppose(m_i) & Space(6)
Next
TextBox3.Text &= vbCrLf
End If
Next
MsgBox("变型空间在特殊假设和一般假设划分出来")
End Sub
End Class
e: \ mydata.txt follows:
Name, shape, taste, kind, generally eat, the size of the
apple, rounded, sweetness, fruit, raw, small, a
watermelon, rounded, sweetness, fruit, raw, medium, it is
bitter, elongated , bitterness, vegetables, cooked food, medium-sized, no
bananas, elongated, sweet, fruit, raw, small, no
pumpkin, round, sweet, vegetables, cooked food, medium-sized, is
lychee, round, sweet flavor, fruit, raw, small, is
Candidates for elimination characteristics of samples collected and grouped, variations derived space, the unknown samples may then be classified method, based on its analysis of the sample on the basis of the attribute set summary, the basic principle is:
1, the establishment of two assumptions, one is generally assumed that a special hypothesis through the analysis of each sample, the expansion of specific hypotheses, reduced general assumptions.
2, assuming the property value is set for each sample, and can be set to a fixed value, can be set to an unknown value? That is, what values are acceptable, if set to nothing, to represent any property values do not accept
3, each line of analysis, the samples are extracted each attribute of each line
1) For a positive example:
you want to delete is generally assumed that no positive cases assumptions consistent with the
need to be specific hypotheses deleted and attribute positive patients did not meet the (ie, to "?"), began special Assuming all is NOTHING, that is not to accept any property values
2) for a counterexample
need to remove the general assumptions in line with the counter-example hypothesis, then add the counterexample assumed containing inconsistent with the attribute value, the other attribute value of "?" in
3) the substance is generally assumed that anti-described characteristics of the embodiment, assuming the special features described n Example
