Model inspection

The command datanova prints an anova table of the data. This anova reflects any location or scale adjustments that have been made to the data. The command resanova prints an anova table for the current set of residuals. The command show prints an anova table for the candidate layer.

The command resid prints the residual matrix in the output window. The resid command is useful for small data sets, but when the matrix is large a better alternative is the command fileres. This will write the residuals as an $n$ by $p$ matrix to a .res file. For example if the data were in MyData.dat the residuals will overwrite the file MyData.res. The previous contents of MyData.res are lost. These files can be very large, and so overwriting them saves disk space. The user should move, copy, or rename any residual files of permanent interest. Such file saving can be done through the operating system in a separate window.

The residual file can then be viewed in an image plot, in some other program than Plaid. Matlab $^{\mbox{\scriptsize TM}}$ or Splus $^{\mbox{\scriptsize TM}}$ are able to make color image plots from flat files of data. Both are able to run concurrently with Plaid.

The command filelayer k writes the $k$'th layer in the model to a .lyr file, overwriting the previous contents. If the data were in MyData.dat the layer will overwrite the file MyData.lyr. Only the rows with $\rho_{ik}=1$ and the columns with $\kappa_{jk}=1$ are written out. Color image plots can be made of these layer files. Layer files of permanent interest must be moved, copied, or renamed.

The command describe prints a description of the model layer by layer. It shows the type, dimensions and sum of squares for each layer. It also shows the sign of $\mu_k$ for each layer.

The commands

compare rows
compare columns
compare data
compare

describe the extent to which the layers in the model overlap. The first three commands each produce a table with one row and one column for each layer. The entry in the table shows the percent of rows (respectively columns or data) appearing in the layer at the left margin that are also in the layer at the top margin. The fourth command produces all three comparisons.

The command print prints a description of all the layers. The description includes the row names, the column names, $\mu_k$, and any applicable $\alpha_{ik}$ and $\beta_{jk}$ values.

The command printa k prints the $\alpha_{ik}$ values if any in layer $k$ along with the names of the rows in layer $k$. The command printb k prints the $\beta_{jk}$ values if any in layer $k$ along with the names of the columns in layer $k$.

The command rowcases prints out three columns for each layer in the model. The first column gives the layer number. The second columns gives the index $i$ between $1$ and $n$ for each row in the layer. The third column gives $\mu_k+\alpha_{ik}$ if the layer has $\alpha_{ik}$ values, and $\mu_k$ otherwise. The point of printing row indices instead of row names is that these values can be read into other software and used to identify the rows directly. The command rowcases k prints out these three columns only for layer $k$. The commands colcases and colcases k print column indices $j$ in layer $k$ and, if applicable, values of $\mu_k+\beta_{jk}$.

The command rowins i prints the data values for row $i$ along with the contribution to this row from each layer containing it, and the present residual for row $i$. The command colins j prints analogous information for column $j$.

The command latexlayer k produces a textual representation of layer $k$ that can, with minor editing, be used as a Table in LaTeX articles. Underscores are removed from the row and column names that appear in the LaTeX table.