Program written, modified, or edited at StatSoft, Inc.}

randomaccess;

{user selects variables to check for outliers}
redim varlist(NVars);
if (SelectVariables1 ("Outlier Analysis:  Please pick the variables", 1, NVars, VarList, count, ?Text1$))=0 then abort;

{user defines the outliers}
coeff := 2;
if (DisplayNumericInputBox
('Please define outliers', 'Outliers are how many standard deviations from the mean?', coeff))=0 then abort;

{Let's user specify where the output is headed}
output_choice := DisplayButtonBox ('Direct output to where?', 'Scrollsheets|Text/Output Window');
if (output_choice = 0) then abort;

{data file is displayed as scrollsheet and new column is added to code for cases with outliers}
if (output_choice = 1) then
	begin
	datatitle$ := 'Marked cases have a value > ' + str(coeff,4,2) + ' std devs from the mean';
	datahandle := MatrixDisplay (Data, datatitle$);
	for n:=1 to NVars do
		ScrollsheetSetColumnName (datahandle, n, VarName(n), ?Name2$);
	ScrollsheetAppendColumns (datahandle, 1);
	ScrollsheetSetColumnName (datahandle, Nvars+1, "OUTLIER?", ?title2$);
	end;

{Big loop so that outlier analysis is performed for all variables selected}
for numvars := 1 to count do
begin

	{read the values from that variable into an array}
	redim varvalues (NCases);
	for counter :=1 to NCases do
		begin
		varvalues(counter) := Data(counter,varlist(numvars));
		end;

	{compute mean and standard deviation}
	ValMean (varvalues, 1, NCases,mean);
	ValStDeviation (varvalues, 1, NCases, std);

	{compute the valid range for the chosen variable given the user's selection of
	the number of standard deviations from the mean}
	low := (mean - (coeff*std));
	high := (mean + (coeff*std));

	{checking for outliers}
	number_outliers := 0;
	for counter2 := 1 to NCases do
		begin
		if ((Data (counter2, varlist(numvars)) < low) or (Data (counter2, varlist(numvars)) > high)) then
			begin
			number_outliers := number_outliers + 1;
			redim outliers(number_outliers);
			outliers(number_outliers) := counter2;
			end;
		end;

	{creates array and stores descriptive info about the variable}
	redim var_info(4);
	var_info(1):=mean;
	var_info(2):=std;
	var_info(3):=high;
	var_info(4):=low;

	{Let's user know if there are no outliers for the specified variable}
	if ((number_outliers = 0) and (output_choice = 1)) then
		DisplayMessageBox (MB_IconExclamation, 'Outlier Analysis', 'No outliers were found for '+
			Varname(varlist(numvars)));

	{Scrollsheet Output}

	{takes care of initial variable}
	desc_title$ := 'Descriptives & Values within ' + str(coeff,4,2) + ' Standard Dev(s) of the Mean';
	if ((output_choice = 1) and (numvars = 1)) then
	begin
	desc_handle := NewScrollsheet (4, 1, var_info, desc_title$,
	 ?rownames$, Varname(varlist(numvars)));
	outlier_title$ := "Cases with Outliers";
	outlier_handle := NewScrollsheet(number_outliers,1, outliers, outlier_title$,?RowNames$,
	Varname(varlist(numvars)));
	ScrollsheetSetRowNameWidth (desc_handle, 20);
	ScrollsheetSetColumnName (outlier_handle, numvars,Varname(varlist(numvars)),'Cases');
	high_temp$ := 'Mean +*s.d.';
	high$ := sinsert(high_temp$,str(coeff,4,2),7);
	low_temp$ := 'Mean -*s.d.';
	low$ := sinsert(low_temp$, str(coeff,4,2),7);
	ScrollsheetSetRowName (desc_handle, 1, 'Mean');
	ScrollsheetSetRowName (desc_handle, 2, 'Standard Deviation');
	ScrollsheetSetRowName (desc_handle, 3, high$);
	ScrollsheetSetRowName (desc_handle, 4, low$);
	ScrollsheetSetColumnFormat (outlier_handle, numvars, SCF_integer, 4);
	for d:=1 to number_outliers do
		begin
		ScrollsheetSetHilite (datahandle, outliers(d), Varlist(numvars), 1);
		ScrollsheetSetValue (datahandle, outliers(d), nvars+1, 1);
		end;
	end;

	{takes care of subsequent variables}
	If ((output_choice = 1) and (not(numvars = 1))) then
	begin
	ScrollsheetAppendColumns (outlier_handle, 1);
	ScrollsheetSetColumnName (outlier_handle, numvars,Varname(varlist(numvars)),'Cases');
	ScrollsheetSetColumnFormat (outlier_handle, numvars, SCF_integer, 4);
	Current_size := ScrollsheetGetNbRows (outlier_handle);
	If (number_outliers > current_size) then
	ScrollsheetAppendRows (outlier_handle, number_outliers - current_size);
	for c:=1 to number_outliers do
		begin
		ScrollsheetSetValue (outlier_handle, c, numvars, outliers(c));
		end;
	for e:=1 to number_outliers do
		begin
		ScrollsheetSetHilite (datahandle, outliers(e), varlist(numvars), 1);
		ScrollsheetSetValue (datahandle, outliers(e), nvars+1, 1);
		end;
	ScrollsheetAppendColumns (desc_handle, 1);
	ScrollsheetSetColumnName (desc_handle, numvars,?Name$ ,Varname(varlist(numvars)) );
	ScrollsheetSetValue (desc_handle, 1, Numvars, mean);
	ScrollsheetSetValue (desc_handle, 2, Numvars, std);
	ScrollsheetSetValue (desc_handle, 3, Numvars, high);
	ScrollsheetSetValue (desc_handle, 4, Numvars, low);
	end;

	{T/O ouput}
	if (output_choice = 2) then
	begin
	Writeln("");
	Writeln ("OUTLIER ANALYSIS OUTPUT");
	Writeln("");
	Writeln("Variable: ",Varname(varlist(numvars)));
	Writeln ("Mean: ", mean, "Standard deviation: ", std);
	Writeln ("Values within +/- ", coeff, "standard deviations are from ", low, "to ", high);
	Writeln("");
	number_outliers := 0;
	for counter2 := 1 to NCases do
		begin
		if ((Data (counter2, varlist(numvars)) < low) or (Data (counter2, varlist(numvars)) > high)) then
			begin
			writeln ("Case number: ", counter2, "(Case Name: ", CaseName(counter2),")", " is an outlier for ",
			varname(varlist(numvars)));
			number_outliers := number_outliers + 1;
			end;
		end;
	if (number_outliers = 0) then
		writeln ("No outliers were found for ", varname(varlist(numvars)));
		end;
end;

{message for scrollsheet output}
if (output_choice = 1) then
DisplayMessageBox (MB_IconExclamation, 'Please note', 'If you selected more than 1 variable, you will need to resize the
scrollsheets to view the output.');

2300 East 14th Street, Tulsa, OK 74104
Phone: (918) 749-1119; Fax: (918) 749-2217
e-mail: info@statsoft.com

©Copyright StatSoft, Inc., 1984-2004.
StatSoft, StatSoft logo, STATISTICA, SEWSS, SEDAS, Data Miner, SEPATH and GTrees are trademarks of StatSoft, Inc.