LAND USE REGRESSION analysis

[Cruise]

Dear SAS experts:

I'd like to assess the spatial association between N_DISEASE and RESP_RISK controlling to minority, income and education using LUR (land use regression) method, as shown in the mock data below. I searched how to solve this modelling problem in SAS with limited success. I will greatly appreciate your direct help in coding or sharing resources with examples for coding?

Thanks in advance. 

My data includes following information and more. 

BLOCK_ID (census block)
BLOCK_POPUL (population)
PCT_MINOR_POP (percent of minority population)
CT_LOWINCOME (percent of low income population)
PCT_LT_HIGHSCHOOL (percent of people with lt high school education)
N_DISEASE (n of health outcomes)
RESP_RISK (absolute value of risk).

DATA HAVE;
INPUT BLOCK_ID BLOCK_POPUL PCT_MINOR_POP PCT_LOWINCOME PCT_LT_HIGHSCHOOL N_DISEASE RESP_RISK;
CARDS; 
60010001001 846 0.78 0.42 0.15 67 205.03 
60010001002 1088 0.88 0.66 0.18 80 333.26 
60010002001 1854 0.95 0.71 0.16 92 805.83 
60010002002 694 0.91 0.37 0.16 61 138.41 
60010002003 1211 0.9 0.41 0.04 15 298.08 
60010002004 672 0.71 0.4 0.06 72 56.48 
60010003001 921 0.35 0.24 0.05 86 69.85 
60010003002 1120 0.78 0.84 0.22 37 382.01 
60010003003 1461 0.21 0.26 0.09 16 21.77 
360010003004 1775 0.8 0.71 0.19 61 462.24 
360010004011 2377 0.13 0.16 0.07 12 89.14 
360010004031 1320 0.4 0.14 0.02 67 82.92 
360010004032 1590 0.36 0.22 0.09 55 51.01 
360010004033 1022 0.27 0.09 0.05 76 167.67 
360010004034 887 0.32 0.31 0.05 62 66.87 
360010004041 5045 0.38 0 0.08 74 895.64 
360010005011 1308 0.62 0.63 0.16 56 275.23 
360010005012 508 0.37 0.33 0.09 29 20.79 
360010005013 483 0.67 0.11 0 69 50.80 
360010005014 1368 0.69 0.46 0.13 16 287.07 
360010005021 2083 0.3 0.66 0 73 92.05 
360010005022 1673 0.34 0.4 0.04 20 102.47 
360010006001 670 0.65 0.81 0.29 82 186.43 
360010006002 1185 0.74 0.74 0.15 66 389.03 
360010006003 1955 0.65 0.76 0.21 85 501.90 
360010007001 867 0.97 0.48 0.17 23 267.87 
360010007002 682 0.89 0.56 0.16 41 159.60 
360010007003 1139 0.94 0.24 0.14 93 175.14 
360010007004 1025 0.95 0.83 0.25 100 414.95 
360010008001 407 1 0.69 0.11 24 183.54 
360010008002 856 0.9 0.69 0.28 81 414.94 
360010008003 774 0.7 0.67 0.24 81 286.64 
360010011001 1239 0.5 0.72 0.18 44 430.26 
360010014001 1752 0.35 0.27 0.07 34 164.59 
360010014002 1168 0.16 0.27 0.02 19 277.72 
360010014003 748 0.64 0.64 0.05 92 127.55 
360010015001 1423 0.27 0.7 0.02 16 30.96 
360010015002 1577 0.32 0.64 0.03 38 15.35 
360010015003 1178 0.55 0.66 0.18 48 220.59 
360010015004 902 0.35 0.29 0 1 114.65 
360010016001 1331 0.27 0.69 0.08 73 13.43 
360010016002 819 0.26 0.11 0.07 73 128.13 
360010016003 837 0.29 0.28 0 57 88.57 
360010016004 1176 0.13 0.17 0 39 249.84 
360010017001 957 0.37 0.39 0.12 94 11.88 
360010017002 1372 0.17 0.31 0.06 25 144.34 
360010017003 1569 0.47 0.29 0.03 27 29.52 
360010017004 664 0.07 0.03 0 64 175.18 
360010018011 860 0.12 0.02 0 15 175.79 
360010018012 1695 0.13 0.18 0.07 26 132.05
;

[PaigeMiller · Accepted solution]

As I understand it, Land Use regression is just a multiple regression, and there are examples of multiple regression in the PROC REG documentation. https://documentation.sas.com/?cdcId=pgmmvacdc&cdcVersion=9.4&docsetId=statug&docsetTarget=statug_re...

[Cruise]

I understand that LUR has multiple regression component. And wondering how GIS comes into an account on the top of that. I understood that LUR helps in predicting the values where exposure was not directly measured using land-use data. I guess land-use data comes from GIS and somehow integrated with multiple regression outputs? Deep learning curve here

[Cruise]

I think I'm confusing myself with a LUR. My exposure data is already estimated at census blocks so is my health outcome. Other studies needed to estimate the exposure at the locations with no measurements which is not my case.