dimensional matrix. A three-dimensional matrix (see below) comprises a series 

 (or set) of tables, where each table ( plane ) contains the same number of rows 

 and columns. All the numbers in a single matrix plane are usually related in 

 some way. 



COLUMNS > 



COLUMNS > 



ROWS 



ROWS 



Matrix 



two-dimensional form 



(table) 



PLANES 



Matrix 

 three-dimensional form 



In the above diagrams, each cell in a sample column vector has been 

 labelled with a "C", each cell of a sample row vector with an "R," and each cell 

 of a plane vector with a "P." 



An individual 

 numbers are, by con 

 tions indicated by 

 with "C" above are 

 always be referred 

 tion by row, column 

 can be described as 

 The three numbers c 

 dimensional, as in 

 simply comprise a _s 

 (2,1,1), where all 

 cell in a single-ce 



row, column, or plane may be referred to by a number, and 

 vention, assigned in order (starting with 1) in the direc- 

 the arrows in the diagrams above. Thus all the cells labeled 

 contained in column "3." An individual cell in a matrix can 

 to by a unigue set of three numbers, one each for its posi- 

 and plane. Thus the locus for the cell labelled "A" above 

 row 1, column 2, and plane 3, or alternatively, "(1,?,3)." 

 an always be assigned, even if the matrix is effectively two- 

 a table, or even one-dimensional (e.g., a "matrix" might 

 ingle cell ). The point "B" above could be located by 

 the cells in a table would be assigned plane number 1. A 

 11 "matrix" would therefore be located at (1,1,1). 



b. Rationale for the MATRIX System 



There are 

 First, many an 

 Easy Graphing, 

 either in row 

 points to be u 

 files contain 

 dispersed thro 

 sis. Data poi 

 raw data files 

 if, over a Ion 

 ferent format) 



two underlying reasons for the de 

 alytical program packages such as 



and the SYMAP spatial mapping sys 

 and column form or in some other s 

 tilized occur together (and seguen 

 data points that, for a certain de 

 ughout the file; they must be "bro 

 nts to be analyzed together might 

 . This dispersion of data points 

 g period of time, many different k 



are collected and entered as comp 



velopment of the MATRIX system. 

 SPSS, BMDP, MINITAB, PLOT-10 

 tem reguire input data that is 

 pecial form in which all data 

 tially). Second, many raw data 

 sired analysis, are in some way 

 ught together" prior to analy- 

 even be scattered over several 

 can be especially troublesome 

 inds of data (each with a dif- 

 uter data files. 



The above conditions result in what could be called a "format gap." There 

 are two aspects of this gap: one is that the raw-data format is not suitable 

 for direct entry of the data into an analytic routine; the other is that data 

 points required for an analysis do not occur together. The MATRIX program 



138 



