Re: Write a code on SAS on likelihood ratio

Arjayita

I have one dataset DORMANT_PCA_CC which has fourteen variables. Bu_419_DQ, Bu_26_ab_num, Bu_345_hn_num, rr_chr_105, bu_162_gg_num, rr_chr_169, bu_799_teb_num, rr_chr_111, bu_301_pl_num, bu_1552_shc_num, bu_1577_ric_num, rr_chr_122, e1_b_07, bu_670_uz_num.
i want to drop one variable which is bu_670_uz_num.

If I want to test whether the drop out variable bu_670_uz_num is significant or not, I shall perform a likelihood ratio test of two models. How to write a code on SAS to check the change in likelihood?

thanks in advance

Arjayita

LIBNAME dev "/team/decision_science/Customer_Modelling/Project/013_Foundation_Customer_Score/Data/Development Sample/Jun19/Stratified Samples" access=readonly;
LIBNAME MRAV "/team/analytics_mod/Technical_Validation/01_Retail/04_Unsecured/FCS_StageD_Rerun/Data";

data dormant_pca_cc;
    set dev.fcs_jun19_dorm_pca_cc_build;
	drop CSV2_2_exclusion_7_bnkrpt CSV2_2_exclusion_8_dceasd CSV2_2_exclusion_9_basic_dorm
		 bcoms_active_9m_frm_Jun19 bcoms_stays_active_9m_frm_Jun19;
run;

%SYSMSTORECLEAR;
LIBNAME modmacro "/team/decision_science/xTeam/.modelling_macro/" access=readonly;
OPTIONS noimplmac obs=max mstored sasmstore=modmacro;

* Create the bindings dataset *;
DATA bandings;
     LENGTH varname $ 40 grps $ 2000;
     INPUT type modtype inmod varname $ grps $;
CARDS;
2 1 1 BU_419_DQ :"0_","U","ND":
1 1 1 BU_26_AB_NUM :0:
1 1 1 BU_345_HN_NUM :0:25:75:
1 1 1 RR_CHR_105 :200:
1 1 1 BU_162_GG_NUM :120:180:360:
1 1 1 RR_CHR_169 :-9999997:5:30:
1 1 1 BU_799_TEB_NUM :0:1:2:
1 1 1 RR_CHR_111 :2000:
1 1 1 BU_301_PL_NUM :0:
1 1 1 BU_1552_SHC_NUM :.N:0:
1 1 1 BU_1577_RIC_NUM :0:50:
1 1 1 RR_CHR_122 :0:
2 1 1 E1_B_07 :"0","N","T","U":
1 1 1 bu_670_uz_num :-1:1000:
;
RUN;

* Create the dataset labels *;
DATA labels;
     LENGTH varname $ 15 grps $ 300;
     INPUT varname $ grps $;
CARDS;
RUN;


* Execute the model SAS macro *;
%MODEL(
     set=dormant_pca_cc,
     scoreset=MRAV.dormant_pca_cc,
     outcome=FCS_bad_9m_Jun19,
     weight=weight,
     holdout=holdout,
     oscore=CSv2_2,
     drop=1,
     user=Arjayita Ghosh,
     progname=Dormant_PCA_CC
);

Arjayita

I have attached the base code for better understanding 🙂

ballardw

A macro call with a parameter list without the complete definition of the macro as in

%MODEL(
     set=dormant_pca_cc,
     scoreset=MRAV.dormant_pca_cc,
     outcome=FCS_bad_9m_Jun19,
     weight=weight,
     holdout=holdout,
     oscore=CSv2_2,
     drop=1,
     user=Arjayita Ghosh,
     progname=Dormant_PCA_CC
);

Does not really provide much for better understanding.

Some where there is code that starts with

%macro Model(<parameters>);
 
and ends with

%mend;

that describes the macro and we could discuss that.

Ksharp

Check @Rick_SAS blogs:
https://blogs.sas.com/content/iml/2024/03/27/likelihood-ratio-test.html

Write a code on SAS on likelihood ratio