NAME

m.ipf - Iterative proportional fitting for error matrices.
(GRASS Data Import/Processing Program)

SYNOPSIS

m.ipf
m.ipf help
m.ipf [-emz] [input=name] [format=string] [stop=value]

DESCRIPTION

m.ipf uses an error or confusion matrix produced by r.coin or r.kappa, smooths zero counts, and does iterative proportional fitting to normalize the matrix.

OPTIONS

Flags:

-e
Indicate when the iterative algorithm finished.
-m
Print the marginals (row and column totals) with each matrix.
-z
Print the intermediate (smoothed) matrix.

Parameters:

input=name
The input file must have the following format: the first line contains an integer K which is the number of rows and columns in the matrix; the remainder of the file is the matrix, i.e., K lines, each containing K integers. If the input is not specified on the command line, it may come from standard input.
format=string
Specifies the format conversion string used to print the results. Default is %7.3f. For details, see the UNIX man page for printf.
stop=value
The stopping criteria is a floating point number which actually specifies an integer maximum number of iterations and a fractional change in marginal. The default, 100.01, specifies that the interative proportional fitting will stop at 100 iterations or when marginals do not change by 0.01, whichever comes first.

EXAMPLE

For the following input,
3
712    0   12
  0  584    2
 18    0  434
zero counts in the matrix will be smoothed:
711.249   0.438  12.314 
  0.443 583.289   2.268 
 18.309   0.273 433.418 
and the matrix will be normalized to yield:
0.969 0.001 0.022 
0.001 0.999 0.004 
0.031 0.001 0.973 

NOTES

Iterative proportional curve fitting is useful when comparing the output of image classification algorithms (for example, i.maxlik and i.smap), especially when training fields (signatures) and/or test fields are different. The diagonals of the normalized matrix can be used in a Tukey multiple comparison test.

SEE ALSO

Assessing Multiple Classifications - GRASS Tutorial on m.ipf

r.coin
r.kappa

Zhuang, X., B.A. Engel, X. Xiong, and C. Johanssen. 1994. Analysis of Classification Results of Remotely Sensed Data and Evaluation of Classification Algorithms, Photogrammetric Engineering and Remote Sensing (in press)

AUTHOR

James Darrell McCauley, Agricultural Engineering, Purdue University