Dissimilarity

One of the types of objects in Praat.

It represents a one-way table with dissimilarities between "objects".

Creating a Dissimilarity from data in a text file

Suppose you have three objects A, B and C. In one way or another, you have acquired the following (symmetric) dissimilarities: δAB = 2 (= δBA) , δAC = 1 (= δCA), and δBC = 1.4 (= δCB), where δAB represents the dissimilarity between object A and object B.

You can create a simple text file like the following:

"ooTextFile" ! The line by which Praat can recognize your file
"Dissimilarity" ! The line that tells Praat about the contents
3 "A" "B" "C" ! Number of columns, and column labels
3 ! Number of rows
"A" 0 2 1 ! Row label (A), A-B value, A-C value
"B" 2 0 1.4 ! Row label (B), B-A value, B-C value
"C" 1 1.4 0 ! Row label (C), C-A value, C-B value

Notice that:

• the row and column labels are identical.
• the matrix elements on the diagonal are zero.
• the matrix is symmetrical.

This text file can be read with the Read from file... command. Since a Dissimilarity object has the data structure of a square symmetrical TableOfReal, you could also start from an appropriate TableOfReal object and cast it to a Dissimilarity object.

Commands

Creation

• Confusion: To Dissimilarity...

Drawing

• Draw as numbers...
• Draw as squares...

Query

• Get column mean (index)...
• Get column mean (label)...
• Get column stdev (index)...
• Get column stdev (label)...
• Dissimilarity: Get additive constant

Modification

• Formula...
• Set value...
• Remove column (index)...
• Insert column (index)...
• Set row label (index)...
• Set row label (label)...
• Set column label (index)...
• Set column label (label)...

Multidimensional scaling analysis

• Dissimilarity: To Configuration (monotone mds)...
• Dissimilarity: To Configuration (i-spline mds)...
• Dissimilarity: To Configuration (interval mds)...
• Dissimilarity: To Configuration (ratio mds)...
• Dissimilarity: To Configuration (absolute mds)...
• Dissimilarity: To Configuration (kruskal)...
Transformations
• Dissimilarity: To Distance...
• Dissimilarity: To Weight

Links to this page


© djmw 20010327