TOOLS > PROFIT
PURPOSE Implements property fitting (ProFit) for a set of co-ordinates and a collection of continuous attributes
DESCRIPTION Given a set of co-ordinates (derived from multidimensional scaling for example) and a contiuous attribute variable this routine uses multiple regression to find the best fit of the projected co-ordinates on to the attribute variable and then plots the information in two dimensional space. In addition it uses a permutation test to give a significance level. It is typically used to test hypothesis about the underlying dimensions in a multidimensional plot with respect to independently measures actor attributes.
Name of file containing 2D coordinates, these should be given with column 1 as the x coordinates and col 2 as the y coordinates. Data type: Valued nx2 matrix.
Name of file containing input attributes. The value of actor i's j attribute should be specified in row i column j. Data type valued matrix
Output coordinates (Default=<inputfilename>-profit)
Name of file containing output coordinates, these consist of the input coordinates together wih one extra coordinate pair for each attribute together with the origin which is labelled +.
Output statistics (Default = ProfitStats)
Name of file containing the regression coefficients, r squareds and p values.
Number of permutations for signifcance test (Default=10000)
Number of random permutations used to generate the p values.
LOG FILE The output first gives a 2D scatterplot of the output co-ordinates, for details of the scatterplot viewer see
scatterplot help file.
The original coordinates are plotted together with the origin labelled with a red + and each of the attributes with their column label highlighted in yellow. The direction cosines are constructed by joining the origin to the attribute coordinate with the direction being from the origin to the attribute.
A table with the regression coefficients the r squared and the p value for each atrtribute.