Page 316 - Excel for Scientists and Engineers: Numerical Methods
P. 316

CHAPTER 13      LINEAR REGRESSION AND CURVE FITTING                  293



                     Type  the  LINEST  formula  with  its  arguments,  in  this  example
                     =LINEST(F3:F14,E3: E14,TRUE,TRUE).    You  can  use  the  following
                     "shorthand"  for  the  logical  arguments  const  and  stats:  FALSE can  be
                     represented  by  0  and  TRUE  by  any  nonzero  value,  as  in  the  formula
                     =LINEST(F3:F14,E3:E14,1,1).
                     Enter the formula by using CONTROL+SHIFT+ENTER.
                   When you "array-enter" a formula, Excel puts braces around the formula, as
               shown below:
                   {=LINEST(F3:F14,E3:E14,1,1)}















                         Figure 13-6.  Regression results and statistics returned by LINEST
                                  for the methane hydrate phase diagram data.
                    (folder 'Chapter 13 Examples', workbook 'Methane Hydrate', sheet 'Phase diagram data')



                   You do not type the braces; if you did, the result would not be recognized by
               Excel as a formula.
                   When  the  LINEST function  is  applied  to  the  data  in  columns  E  and  F  of
               Figure 13-2, the results shown in Figure 13-6 are obtained.
                   As  you  can  see,  LINEST  returns  a  large  amount  of  useful  statistical
               information simply by entering a single formula: the regression coefficients, their
               standard  deviations,  the R2 value,  plus  several other statistical  quantities.  You
               must,  however,  be  familiar  with  the  layout  of regression  results  and  statistics
               shown  in  Figure  13-5 (also  shown  in  Excel's  On-Line  Help  for  the  LINEST
               worksheet function) in order to know what value each cell contains.


                Multiple Linear Regression Using LINEST
                   Now  that we've  gained  some  familiarity with  LINEST,  let's  apply  it  to  an
               example  of multiple  linear  regression.  The data table  in  Figure  13-7 lists the
                freezing points of solutions of ethylene glycol.  We want to be able to obtain the
                freezing  point  of  a  solution  of ethylene  glycol  with  wt% that  is  intermediate
                between the data values given in the table.
   311   312   313   314   315   316   317   318   319   320   321