HISTO <data>,<variable>,L GHISTO <data>,<variable>,L / for graphic screen forms the frequency distribution and plots the histogram of <variable> in <data> according to a classification given by <variable>=<lower limit>(<step>)<upper limit> L (optional parameter) is the first line for the table of the frequency distribution. The results are saved also in an output file, if that is specified (see OUTPUT). Furthermore the frequency distribution is saved in file FREQ.F on the data disk and it can be used later by a HISTO operation of the form HISTO data>F,<variable>,L. Especially histograms of large data files are replotted faster by this option thus permitting fitting of various theoretical distributions to the same frequency distribution (see FIT below). Example: HISTO DATA1,Weight / Weight=40(2)100 Classes of Weight will be 1: 40<Weight<=42 2: 42<Weight<=44 ------- 30: 98<Weight<=100 Various theoretical distributions may be fitted and the corresponding frequency curve is plotted on the histogram by using a specification of the form FIT=<name of the distribution>. The following distributions are available: NORMAL(mean,variance) X is N(mean,variance) LOGNORMAL(mean,variance) log X is N(mean,variance) UNIFORM(a,b) X has uniform distribution on (a,b) BINOMIAL(N,p) X is Bin(N,p) POISSON(mean) X is Poisson(mean) The user may also define his/her own distribution (see below). If the parameters of the distribution are not given, the maximum likelihood estimates computed from the frequency distribution are used. Examples: FIT=NORMAL(175,36.5) FIT=NORMAL If FIT is used, the results will also include the fitted frequencies and the common Chi-square test. In this test consecutive classes are com- bined when necessary to meet the condition expected frequency >= min where <min> is given by MINF=<min>. Default is MINF=5. When using FIT for discrete distributions the <variable> specification must be of the form <variable>=-0.5(1)<max.value+0.5>. In continuous distributions, theoretical frequencies are computed by numerical integration using a step length <step>/2. To replace this by <step>/K, a specification STEPDIVISOR=K must be given. Other extra specifications of HISTO are HOME, SIZE, XDIV, YDIV, PEN, LINETYPE, XSCALE, YSCALE, FRAME, GRID, TICK, HEADER, XLABEL, YLABEL, TEXT, IND, VALUES being mostly the same as those of the PLOT operation for scatter diagrams. The bars in the histogram are filled with the shading [FILL1] defined in the current plotting driver. Redefinition like [FILL1]=[FILL4] is allowed in the HISTO scheme. By FILL=NO bars are not filled at all. In PostScript plotting the color for bars can be selected, for example, by setting [FILL-1]=1,1,0.5,0 (cmyk components) FILL=-1 . By default, an error message is given if no observations are accepted. This message is removed by ACCEPT=0. Also an error message appears when all observations belong to the same class. This message is removed by ACCEPT=1 (or ACCEPT=0). U = User-defined distributions for HISTO S = Information on scatter diagrams