When you use a Binary Logistic Regression Model to estimate the probability of an event you can then set a threshold value for this probability so that if P > K then classify the case as Event = 1 otherwise classify the case as Event = 0   There are four possible decisions :   Classify the case as Event = 0  when it is in fact Event = 0 Classify the case as Event = 0  when it is in fact Event = 1     TYPE 2 ERROR  [ Cost of False Negative Error = X ] Classify the case as Event = 1  when it is in fact Event = 0     TYPE 1 ERROR  [ Cost of False Positive   Error = Y ] Classify the case as Event = 1  when it is in fact Event = 1   Question:   If I can estimate the values of the Costs of making either a Type 1 Error or  a Type 2 Error before I start to build the Logistic Regression Model, how can I include this information in the Proc Logistic Code in order to optimise my model with respect to  the Error Costs.   Without this inclusion, the default is to assume that the costs are all the same, which in my project is clearly not the case.    I would appreciate it if you could provide guidance w.r.t. SAS Procedures, Documentation , SAS Code as well as Published Papers / Case Studies.   Thanks    ... View more

Given the following SAS Code which is applied to a SAS Date Set containing several explanatory variables and specifically a CHAR Variable: MAIN_DOB which contains the date of birth of the main applicant in the following format: 1978-06-28   We need to calculate the age of the applicant in years at the time of contract initiation.   Contract_Start_Date_DT is a SAS Date in the format 25/06/2014   The following SAS Code works but surely there is a more elegant / efficient solution  to this simple problem   DATA LIBXYZ.DEVELOPMENT_SAMPLE_03 ; SET LIBXYZ.DEVELOPMENT_SAMPLE_02 ; SearchDate_DT = DATEPART(SearchDate) ; QuoteDate_DT = DATEPART(QuoteDate) ; Contract_StartDate_DT = DATEPART(Contract_StartDate) ; Vehicle_RegistrationDate_DT = DATEPART(Vehicle_RegistrationDate) ; Termination_Date_DT = DATEPART(Termination_Date) ; Termination_CarDisposalDate_DT = DATEPART(Termination_CarDisposalDate) ;   DateOfBirth_Char = INPUT( COMPRESS( MAIN_DOB,"-" ) , $8. ) ; DateOfBirth_Year = INPUT( SUBSTR(DateOfBirth_Char , 1 , 4 ) , 4. ) ; DateOfBirth_Month = INPUT( SUBSTR(DateOfBirth_Char , 5 , 2 ) , 2. ) ;   IF DateOfBirth_Month < 10 THEN DateOfBirth_Month1 = CAT('0' , DateOfBirth_Month ) ; ELSE DateOfBirth_Month1 = DateOfBirth_Month ; DateOfBirth_Day = INPUT( SUBSTR(DateOfBirth_Char , 7 , 2 ) , 2. ) ;   IF DateOfBirth_Day < 10 THEN DateOfBirth_Day1 = CAT('0' , DateOfBirth_Day ) ; ELSE DateOfBirth_Day1 = DateOfBirth_Day ; DateOfBirth = INPUT(CATX( '/' , DateOfBirth_Day1 , DateOfBirth_Month1 , DateOfBirth_Year ) , ddmmyy10. ) ;   FORMAT SearchDate_DT QuoteDate_DT Contract_StartDate_DT Vehicle_RegistrationDate_DT Termination_Date_DT Termination_CarDisposalDate_DT ddmmyy10. ; FORMAT DateOfBirth ddmmyy10. ; CONTRACT_AGE = ( YEAR(Contract_StartDate_DT) - DateOfBirth_Year ) ; FORMAT CONTRACT_AGE 4. ;   DROP SearchDate QuoteDate Contract_StartDate Vehicle_RegistrationDate Termination_Date Termination_CarDisposalDate DateOfBirth_Char DateOfBirth_Year DateOfBirth_Month DateOfBirth_Day DateOfBirth_Day1 DateOfBirth_Month1 ; RUN;   Thanks for kind assistance ... View more

My SAS Code below is used to    a. Drop Variables from a Data Set b. Construct Macro Variables containing Lists of Variable Names   How can I drop variables from Data Set A based on their membership of Data Set B   /* CONSTRUCT THE MACRO VARIABLE NUM_VARS_LIST TO CONTAIN THE LIST OF NUMERIC PREDICTOR VARIABLES FOR FUTURE ANALYSIS */   ODS OUTPUT NLEVELS = LIBXYZ.NUM_VAR_LEVELS;   PROC FREQ DATA = LIBXYZ.DEVELOPMENT_SAMPLE_03N           /* SOURCE INPUT DATA SET */ NLEVELS ; TABLES _NUMERIC_ / NOPRINT ;                              /* ONLY NUMERIC VARIABLES */ RUN;   /* IDENTIFY THE NON-VARIABLES TO BE DELETED FROM THE INPUT DATA SETS : SELECT THE NUMERIC VARIABLES TO BE DROPPED: */   DATA LIBXYZ.DROPNUMVARS (RENAME = (TABLEVAR = NUMVAR_NAME)); SET LIBXYZ.NUM_VAR_LEVELS (WHERE = (NLEVELS = 1 OR NNONMISSLEVELS IN (0 , 1) )); RUN;   /* There were 2 observations read from the data set LIBXYZ.NUM_VAR_LEVELS WHERE (NLEVELS=1) or NNONMISSLEVELS in (0, 1)  The data set LIBXYZ.DROPNUMVARS has 2 observations and 5 variables. */   TITLE ' NUMERIC VARIABLES WHERE [ NLEVELS = 1 OR NNONMISSLEVELS IN (0 , 1) ] TO BE DROPPED FROM THE DATA SET' ; PROC PRINT DATA = LIBXYZ.DROPNUMVARS; RUN ;   /* DROP THE SELECTED NUMERIC VARIABLES */   How can I drop variables from Data Set LIBXYZ.DEVELOPMENT_SAMPLE_03N based on their membership of Data Set LIBXYZ.DROPNUMVARS   After that I want to create the the macro variable list:   PROC SQL NOPRINT; SELECT TABLEVAR INTO :NUM_VARS_LIST SEPARATED BY ' ' FROM LIBXYZ.NUM_VAR_LEVELS                       /* ODS DATA SET NLEVELS = NUM_VAR_LEVELS */ ; QUIT;   /********************************************************************************************************************************/   Thanks for your help. ... View more

PROC SQL;      CREATE TABLE LIBXYZ.DEVELOPMENT_SAMPLE_03      AS      SELECT      T1.CONTRACTNO ,      T1.FLAG                                                                    AS BIN_TARGET , /* BINARY TARGET */       T1.DELPHI_ADDRESS_LINK_NDLNK01_ADDR      AS VAR_0001 ,      T1.DELPHI_APACCS_CCDATASUPPLIED_APA        AS VAR_0002 ,      T1.DELPHI_CAIS_E1B03_WORST_STATUS_L        AS VAR_0003 ,      T1.DELPHI_CAIS_E1B05_WORST_STATUS_L        AS VAR_0004 ,      T1.DELPHI_CAIS_E1B07_WORST_STATUS_L        AS VAR_0005 ,          T1.SPB113                                                                 AS VAR_0267    FROM             LIBXYZ.DEVELOPMENT_SAMPLE_02   T1    ORDER BY    T1.CONTRACTNO;   QUIT;   The renaming technique in the above SQL code is very manual and time-consuming as well as being "hard-coded" and therefore difficult to make consistent changes   I have a data set containing several hundred variables with very long names ( lots with names greater than 32 characters and in some cases the first 32 characters are identical across distinct variables )   I would like to rename all these variables : Var_Long_Name_01  to Var_0001 as indicated in the SQL Code above.   Is there some simple way in which I can do this without having to type in the new name for each of the several hundred variables?   The solution can involve base SAS, SQL or even Excel !!   Thanks for your help ... View more

Using SAS 9.4   I need to split my data set into two subsets   a. All Numeric Variables b. All Character variables   How can I use PROC SQL to Select all the NUMERIC Columns in a Table?   Alternatively how would I do this in Base SAS?   Thanks ... View more