VB.NET中如何为NDtw构造SeriesVariable处理多变量数据

NDtw处理多变量数据时，SeriesVariable要求的是「按时间点组织」的结构：每个元素对应一个时间点的所有变量测量值，而非按变量分组的二维数组。你的原始二维数组是「变量为行、时间点为列」，需要调整为「时间点为元素、每个元素包含对应所有变量值」的格式。

正确的转换方法 #

1. 遍历每个时间点，提取对应所有变量的测量值，组成单个Double()数组
2. 将这些数组集合转换为SeriesVariable类型

修改后的代码 #

Imports NDtw
Imports System.Collections.Generic

Public Class Form1
    Private Sub Button1_Click(sender As Object, e As EventArgs) Handles Button1.Click
        Dim dist As Double
        Dim actualdist As Double

        ' 原始数据：行=变量，列=时间点
        Dim rawSeries1(,) = {{1.0, 3.0, 7.0, 8.0, 11.0, 6.0, 6.0},
                             {3.0, 6.0, 3.0, 7.0, 13.0, 12.0, 7.0}}
        Dim rawSeries2(,) = {{3.0, 6.0, 3.0, 7.0, 13.0, 12.0, 7.0},
                             {1.0, 3.0, 7.0, 8.0, 11.0, 6.0, 6.0}}

        ' 转换为SeriesVariable格式：每个元素是一个时间点的所有变量值
        Dim series1 As New SeriesVariable(ConvertToTimePointSeries(rawSeries1))
        Dim series2 As New SeriesVariable(ConvertToTimePointSeries(rawSeries2))

        Dim dtw As New Dtw(series1, series2)
        dist = dtw.GetCost()

        TextBox3.Text = dist.ToString
    End Sub

    ' 辅助方法：将「变量为行」的二维数组转为「时间点为元素」的数组集合
    Private Function ConvertToTimePointSeries(rawData(,) As Double) As IEnumerable(Of Double())
        Dim timePointCount As Integer = rawData.GetLength(1) ' 时间点数量（列数）
        Dim variableCount As Integer = rawData.GetLength(0) ' 变量数量（行数）
        Dim timePointList As New List(Of Double())

        For timePointIndex As Integer = 0 To timePointCount - 1
            Dim variablesAtTimePoint(variableCount - 1) As Double
            For variableIndex As Integer = 0 To variableCount - 1
                variablesAtTimePoint(variableIndex) = rawData(variableIndex, timePointIndex)
            Next
            timePointList.Add(variablesAtTimePoint)
        Next

        Return timePointList
    End Function
End Class

关键说明 #

• SeriesVariable的构造函数接受IEnumerable(Of Double())类型参数，每个Double()对应一个时间点的多变量观测值
• 辅助方法ConvertToTimePointSeries负责将原始的「变量优先」二维数组，转换为NDtw要求的「时间点优先」结构
• 如果你不想写辅助方法，也可以手动逐个创建时间点数组并添加到SeriesVariable中，比如：

  Dim series1 As New SeriesVariable()
  series1.Add(New Double() {1.0, 3.0}) ' 第一个时间点的两个变量值
  series1.Add(New Double() {3.0, 6.0}) ' 第二个时间点的两个变量值
  ' 依次添加剩余5个时间点的数据
  

内容的提问来源于stack exchange，提问作者Jamie