Predictive Model and Determinants of Under-5 Mortality
Fikrewold Bitew
February 18, 2019
# Load EDHS 2011
library(haven)
library(survival)
library(survey)
library(eha)
library(survminer)
library(car)
library(rgdal)
library(foreign)
library(muhaz)
library(leaflet)
library(sf)
library(sp)
library(raster)
library(rgeos)
library(mapview)Used data from Ethiopan Demographic and Health Survey (EDHS - 2016)
# CHILDREN DHS DATA
dhs_2016<-read.dta("~/Google Drive/DHS Resources/ET_2016_DHS_08152017_2025_103714/etkr70dt/ETKR70FL.dta", convert.factors = F)
edhs16<-subset(dhs_2016,select=c("caseid","b0", "b1", "b2", "b2", "b3", "b4", "b5", "b7","m1", "m15", "m10", "m14", "m70","v005", "v008", "v021", "v022", "v024", "v025","v012", "v013", "h0", "v190", "v201", "v208", "v106", "v3a02", "v113", "m15", "b11", "bord", "v113", "v115", "h10", "h10", "h2", "h33", "v116","v130", "v137", "v212", "v364", "v025", "v445", "v501","v624", "v701", "v445" ))
edhs16$yr<-2016
#WOMEN DHS DATA
#wdhs_2016<-read.dta("~/Google Drive/DHS Resources/ET_2016_DHS_08152017_2025_103714/etir70dt/ETIR70FL.dta", convert.factors = F)
ethsp4<-readOGR(dsn="~/Google Drive/fikre/ethiopia_dhs/sdr_subnational_data_2017-03-28/shps",layer = "sdr_subnational_data_dhs_2016")## OGR data source with driver: ESRI Shapefile
## Source: "/Users/fikrewoldbitew/Google Drive/fikre/ethiopia_dhs/sdr_subnational_data_2017-03-28/shps", layer: "sdr_subnational_data_dhs_2016"
## with 11 features
## It has 55 fields#write.csv(ethsp4, "/Users/fikrewoldbitew/Google Drive/Data/dt.csv")
edhs16$REGION<- Recode(edhs16$v024, recodes='11="Tigray"; 2="Afar"; 3= "Amhara"; 8= "Oromia"; 10="Somali"; 4="Benishangul"; 9="SNNPR"; 6="Gambela"; 7="Harari";1="Addis Adaba"; 5="Dire Dawa";else=NA', as.factor = F)
edhs16<-merge(edhs16, ethsp4@data, by.x="REGION", by.y="REGNAME" ,all.x=T)## identify the working variables and the all the edhs datasets
edhs16$b5<-ifelse(edhs16$b2>=2004,edhs16$b5,NA)
edhs16$under_5<-ifelse(edhs16$b5==0,0,1)
#write.csv(edhs16, "/Users/fikrewoldbitew/Google Drive/PhD_Courses/Spring 2019/before16.csv")
edhs16$under_5[is.na(edhs16$under_5)]<-1
edhs16$chmort<-ifelse(edhs16$under_5==0,1,0)
#write.csv(edhs16, "~/Google Drive/PhD_Courses/after17.csv"Survival Analysis Model Moels
##
## Alive at 5 Dead by 5
## 10006 635##
## Alive at 5 Dead by 5
## 10006 635#Used variables in the Model
edhs16$region<- Recode(edhs16$v024, recodes='11="Tigray"; 2="Afar"; 3= "Amhara"; 8= "1Oromia"; 10="Somali"; 4="Ben-Gumuz"; 9="SNNP"; 6="Gambella"; 7="Harari";1="Addis Ababa"; 5="Dire Dawa";else=NA', as.factor = F)
edhs16$child_sex<-ifelse(edhs16$b4,"Male", "Female")
edhs16$child_sex<-car::Recode(edhs16$b4, recodes='1="Male"; 2="Female"; else=NA')
#edhs16$sibling_alive<-recode(edhs16$mm2_01, recodes='0="Dead"; 1="Alive"; else=NA', as.factor.result = T)
#Education
edhs16$educ<-Recode(edhs16$v106, recodes="0='0Noeducation'; 1='1Primary'; 2:3='2Secodary and Higher'; else=NA", as.factor=F)
edhs16$bmi<-ifelse(edhs16$v445==9999 | edhs16$v445==9998 , NA, edhs16$v445/100)
edhs16$age<-edhs16$v012
edhs16$mother_age<-Recode(edhs16$v013,recodes="1='15-19'; 2:3='20-34'; 4:7='35-49';else=NA", as.factor=T)
edhs16$births_5ys<-edhs16$v208
edhs16$ch_5<-Recode(edhs16$v137, recodes="0:1='0-1'; 2:18='2+';else=NA", as.factor=F)
edhs16$b_order<-Recode(edhs16$bord, recodes="1='1st-2nd'; 2='1st-2nd'; 3:14='3rd or later';else=NA", as.factor=T)
edhs16$b_interval<-Recode(edhs16$b11, recodes="7:23='< 2 years'; 24:48='2-4 yrs'; 48:215='> 4 yrs';else=NA", as.factor=T)
#wealth Index
edhs16$wealthindex<-Recode(edhs16$v190, recodes="1:2='1Poor'; 3='2Middle'; 4:5='3Rich'; else=NA", as.factor=F)
edhs16$religion<-Recode(edhs16$v130, recodes="1='1Orthodoxr'; 2='Others'; 3='3Protestant'; 4='2Muslim';5='Others';else=NA", as.factor=F)
edhs16$ceb<-edhs16$v201
edhs16$contra_use<-Recode(edhs16$v364, recodes=
"1='1Using modern method';
2='4Does not intend to use';
3='3Non-user - intends to use later';
4='4Does not intend to use';else=NA",
as.factor=F)
edhs16$contraceptive_use<-Recode(edhs16$v364, recodes="1='0Yes'; 2:5='No'; else=NA", as.factor=T)
edhs16$water_source<-Recode(edhs16$v113, recodes="10:14='Improved';
20='Improved';
21= 'Improved';
30='Unimproved';
31='Improved';
32='Unimproved';
40='Unimproved';
41='Improved';
42='Unimproved';
43='Unimproved';
51='Improved';
61='Unimproved';
62='Unimproved';
71='Improved';
96='Unimproved' ; else=NA", as.factor = F)
edhs16$time_water<-edhs16$v115
edhs16$toilet_facility<-Recode(edhs16$v116,
recodes='10 = "Improved";
11 = "Improved";
12= "Improved";
13 = "Improved";
14= "Improved";
15= "Improved";
20= "Improved";
21= "Improved";
22= "Improved";
23= "Unimproved";
30= "Unimproved";
31= "Unimproved";
41= "Unimproved";
42= "Unimproved";
43= "Unimproved";
96= "Unimproved";
else=NA', as.factor = F)
edhs16$residence<-Recode(edhs16$v025, recodes='1="Urban"; 2="Rural"; else=NA', as.factor = F)
edhs16$watersource<-Recode(edhs16$v113, recodes="10:13='1Piped/bottled'; 20:32= '2Rain/well/surface/tanker'; 61:62= '2Rain/well/surface/tanker'; 40:43='3River/stream/lake/spring'; 51='2Rain/well/surface/tanker'; 71='1Piped/bottled'; else=NA", as.factor = F)
#Age at first Birth
edhs16$age_fb<-Recode(edhs16$v212, recodes="10:19='<20 years'; 20:49='20+years'; else=NA", as.factor=T)
edhs16$bmi<-ifelse(edhs16$v445==9999 | edhs16$v445==9998 , NA, edhs16$v445/100)
edhs16$bmi_rec<-Recode(edhs16$bmi, recodes="0:18.49='Underweight'; 18.5:25='1Normal'; 25:54='Overweight';
else=NA", as.factor=T)
edhs16$fac_CS_delivery<-Recode(edhs16$m15, recodes="11:12='Home'; 22:25='Fac with CS delivery';
32:35='Fac with CS delivery'; 21='Fac without CS delivery';
31='Fac without CS delivery';26='Fac without CS delivery';
36='Fac without CS delivery';41:46='Fac without CS delivery';
96='Fac without CS delivery';else=NA",
as.factor=T)
edhs16$child_wanted<-Recode(edhs16$m10, recodes="1='1Then'; 2='2Later'; 3='3Not at all'; else=NA", as.factor=T)
edhs16$ever_vaccin<-Recode(edhs16$h10, recodes="0='No'; 1='Yes'; else=NA", as.factor=T)
edhs16$bcg_vaccin<-Recode(edhs16$h2, recodes="0='No'; 1:3='Yes'; else=NA", as.factor=T)
edhs16$vitA_vaccin<-Recode(edhs16$h33, recodes="0='No'; 1:3='Yes'; else=NA", as.factor=T)
edhs16$Postnatal<-Recode(edhs16$m70, recodes="0='No'; 1='Yes'; else=NA", as.factor=T)
edhs16$antenatal_visits<-Recode(edhs16$m14, recodes="0='0_No Visit'; 1:4='1-4 visits'; 5:98='5+ Visits'; else=NA", as.factor=T)Design for Total (Rural+Urban)
# Calculate the DHS sample weight
##remove all missing the BMI parameter
edhs16$wt<-edhs16$v005/1000000
options(survey.lonely.psu = "adjust")
edhs16$yrstratum<-paste(edhs16$SVYYEAR, edhs16$v022, sep="-")
edhs16$yrpsu<-paste(edhs16$SVYYEAR, edhs16$v021, sep="-")
table(edhs16$yrstratum)##
## 2016-1 2016-10 2016-11 2016-12 2016-13 2016-14 2016-15 2016-16 2016-17
## 169 577 488 301 1204 50 829 95 500
## 2016-18 2016-19 2016-2 2016-20 2016-21 2016-22 2016-23 2016-24 2016-25
## 682 175 864 539 201 404 461 264 283
## 2016-3 2016-4 2016-5 2016-6 2016-7 2016-8 2016-9
## 94 968 89 507 381 75 441des<-svydesign(ids=~yrpsu, strata=~yrstratum, weights=~wt, data = edhs16[is.na(edhs16$wt)==F,], nest=T)## Calculate the proportion of Under-5 Mortality in the Design
prop.table(cat<-svytable(~age_fb+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## age_fb 2016
## <20 years 63.96452
## 20+years 36.03548prop.table(cat<-svytable(~v013+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## v013 2016
## 1 3.429153
## 2 18.759059
## 3 30.420495
## 4 22.583805
## 5 16.076227
## 6 6.558483
## 7 2.172777prop.table(cat<-svytable(~births_5ys+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## births_5ys 2016
## 1 42.01802559
## 2 45.75616270
## 3 10.87498219
## 4 1.29207322
## 5 0.05875629prop.table(cat<-svytable(~child_sex+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## child_sex 2016
## Female 48.0625
## Male 51.9375prop.table(cat<-svytable(~residence+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## residence 2016
## Rural 88.97224
## Urban 11.02776prop.table(cat<-svytable(~educ+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## educ 2016
## 0Noeducation 66.081299
## 1Primary 26.769240
## 2Secodary and Higher 7.149462prop.table(cat<-svytable(~residence+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## residence 2016
## Rural 88.97224
## Urban 11.02776prop.table(cat<-svytable(~wealthindex+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## wealthindex 2016
## 1Poor 46.77622
## 2Middle 20.68442
## 3Rich 32.53937prop.table(cat<-svytable(~fac_CS_delivery+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## fac_CS_delivery 2016
## Fac with CS delivery 18.470333
## Fac without CS delivery 8.977616
## Home 72.552051prop.table(cat<-svytable(~contraceptive_use+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## contraceptive_use 2016
## 0Yes 31.29328
## No 68.70672prop.table(cat<-svytable(~child_wanted+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## child_wanted 2016
## 1Then 75.111813
## 2Later 16.747325
## 3Not at all 8.140862prop.table(cat<-svytable(~antenatal_visits+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## antenatal_visits 2016
## 0_No Visit 37.13255
## 1-4 visits 46.58585
## 5+ Visits 16.28160prop.table(cat<-svytable(~Postnatal+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## Postnatal 2016
## No 91.665432
## Yes 8.334568prop.table(cat<-svytable(~region+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## region 2016
## 1Oromia 0.2439650
## Addis Ababa 6.4945668
## Afar 1.0365847
## Amhara 18.8002391
## Ben-Gumuz 44.0083893
## Dire Dawa 4.6080560
## Gambella 1.1035299
## Harari 20.8315225
## SNNP 0.2343264
## Somali 2.2131130
## Tigray 0.4257073prop.table(cat<-svytable(~bmi_rec+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## bmi_rec 2016
## 1Normal 74.159151
## Overweight 6.280034
## Underweight 19.560815prop.table(cat<-svytable(~chmort+SVYYEAR,design=des), margin = 2)*100## SVYYEAR
## chmort 2016
## 0 94.902101
## 1 5.097899reg.est<-svyby(~chmort,~REGION+SVYYEAR,FUN = svymean, na.rm=T, design =des)
reg.est.wide<-reshape(reg.est, idvar = "REGION", timevar = "SVYYEAR", direction = "wide")
reg.est.wide## REGION chmort.2016 se.2016
## Addis Adaba.2016 Addis Adaba 0.03326519 0.005740422
## Afar.2016 Afar 0.08346723 0.013123181
## Amhara.2016 Amhara 0.04706449 0.007561748
## Benishangul.2016 Benishangul 0.05414081 0.006569638
## Dire Dawa.2016 Dire Dawa 0.05970065 0.006738315
## Gambela.2016 Gambela 0.06254413 0.007546533
## Harari.2016 Harari 0.05105880 0.008868284
## Oromia.2016 Oromia 0.05390933 0.009321645
## SNNPR.2016 SNNPR 0.06140410 0.011441805
## Somali.2016 Somali 0.03070644 0.008126967
## Tigray.2016 Tigray 0.05780495 0.020216925
popch = paste0("<strong>", eth.geo$REGNAME, "</strong>", "</br>", eth.geo$CMECMRCU5M)
bins <- c(30,40, 50,60, 70,80, 90,100, 110, 130)
bin <- c(3, 5, 10, 15, 20, 25,30,40, 50,60,70, 100, Inf)
pall <- colorBin("YlOrRd", domain = eth.geo$CMECMRCU5M, bins = bins)
eth.geo$pch_mort<-eth.geo$CMECMRCU5M*100
mypal <- colorQuantile(palette = "YlOrRd", domain = eth.geo$CMECMRCU5M, n = 5, reverse = FALSE)
U5M <- leaflet(eth.geo) %>%
addProviderTiles("OpenStreetMap.Mapnik") %>%
#setView(lat = 9, lng = 41, zoom = 6.2) %>%
setView(lat = 9.292, lng = 40.71 , zoom = 6.2) %>%
addProviderTiles("CartoDB.Positron") %>%
addPolygons( data = eth.geo, fillColor = ~pall(eth.geo$CMECMRCU5M),
fillOpacity = 1,
weight = 1,
color = "white",
popup = popch)%>%
addLabelOnlyMarkers(data = eth.geo, lng = ~X1, lat = ~X2, label = ~REGNAME,
labelOptions = labelOptions(noHide = TRUE, direction = 'center', textOnly = TRUE)) %>%
addLegend(pal = pall, values = ~CMECMRCU5M, title = "U5 Mortality Rate",
opacity = 0.5, position = "bottomright")
U5Mmapshot(U5M, file = "~/Google Drive/PhD_Courses/Spring 2019/PublicHealth/FinalProject/Maps/U5M.png")## PhantomJS not found. You can install it with webshot::install_phantomjs(). If it is installed, please make sure the phantomjs executable can be found via the PATH variable.Fit the GLM model for predicting Under-5 Mortality
#Fit Logistic Regression Model for Under-5 Mortality
#Demographic factors
library(lme4)
library(lmerTest)
require(MASS)
library(pROC)
#demographic Factors
#Demographic Factors and Socio-economic Factors
fit<-glm(chmort~age_fb+child_sex+b_order+b_interval+time_water+water_source+toilet_facility+births_5ys+residence+educ+wealthindex+contraceptive_use+region+bmi_rec+fac_CS_delivery+antenatal_visits+Postnatal+child_wanted,family=binomial(link = "logit"), data=train.m, weights = wt)
summary(fit)##
## Call:
## glm(formula = chmort ~ age_fb + child_sex + b_order + b_interval +
## time_water + water_source + toilet_facility + births_5ys +
## residence + educ + wealthindex + contraceptive_use + region +
## bmi_rec + fac_CS_delivery + antenatal_visits + Postnatal +
## child_wanted, family = binomial(link = "logit"), data = train.m,
## weights = wt)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.7221 -0.2993 -0.1406 -0.0680 5.6020
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) -3.8725243 1.6859802 -2.297
## age_fb20+years -0.4809285 0.1931829 -2.489
## child_sexMale 0.6582764 0.1722778 3.821
## b_order3rd or later 0.4466072 0.2814350 1.587
## b_interval> 4 yrs -0.8167354 0.2838838 -2.877
## b_interval2-4 yrs -0.8283873 0.2033067 -4.075
## time_water -0.0005096 0.0004223 -1.207
## water_sourceUnimproved -0.5895874 0.1811946 -3.254
## toilet_facilityUnimproved 0.4529802 0.4155196 1.090
## births_5ys 0.1838833 0.1423564 1.292
## residenceUrban -0.2075049 0.4708196 -0.441
## educ1Primary -0.2382169 0.2224136 -1.071
## educ2Secodary and Higher 0.5614059 0.4767246 1.178
## wealthindex2Middle -0.0598605 0.2197083 -0.272
## wealthindex3Rich 0.0924289 0.2122506 0.435
## contraceptive_useNo 0.4823318 0.2056633 2.345
## regionAddis Ababa -0.1742372 1.5852972 -0.110
## regionAfar -0.5213084 1.8084809 -0.288
## regionAmhara -0.3179768 1.5579480 -0.204
## regionBen-Gumuz 0.2332658 1.5456120 0.151
## regionDire Dawa -0.3433115 1.5962251 -0.215
## regionGambella -0.2642134 1.7697858 -0.149
## regionHarari 0.1732624 1.5509654 0.112
## regionSNNP -0.1348778 2.4121574 -0.056
## regionSomali 0.2096552 1.8048188 0.116
## regionTigray -0.0822718 2.2897300 -0.036
## bmi_recOverweight -0.4126576 0.4371340 -0.944
## bmi_recUnderweight 0.1085875 0.1964976 0.553
## fac_CS_deliveryFac without CS delivery 1.0613701 0.3794836 2.797
## fac_CS_deliveryHome 0.1833479 0.2748270 0.667
## antenatal_visits1-4 visits -0.4555848 0.1886194 -2.415
## antenatal_visits5+ Visits -0.7835006 0.3149680 -2.488
## PostnatalYes -0.9087024 0.4936503 -1.841
## child_wanted2Later -0.3051008 0.2435738 -1.253
## child_wanted3Not at all 0.3404207 0.2283477 1.491
## Pr(>|z|)
## (Intercept) 0.021625 *
## age_fb20+years 0.012792 *
## child_sexMale 0.000133 ***
## b_order3rd or later 0.112537
## b_interval> 4 yrs 0.004015 **
## b_interval2-4 yrs 4.61e-05 ***
## time_water 0.227472
## water_sourceUnimproved 0.001138 **
## toilet_facilityUnimproved 0.275645
## births_5ys 0.196457
## residenceUrban 0.659408
## educ1Primary 0.284145
## educ2Secodary and Higher 0.238943
## wealthindex2Middle 0.785272
## wealthindex3Rich 0.663221
## contraceptive_useNo 0.019014 *
## regionAddis Ababa 0.912482
## regionAfar 0.773150
## regionAmhara 0.838276
## regionBen-Gumuz 0.880038
## regionDire Dawa 0.829707
## regionGambella 0.881324
## regionHarari 0.911051
## regionSNNP 0.955409
## regionSomali 0.907522
## regionTigray 0.971338
## bmi_recOverweight 0.345166
## bmi_recUnderweight 0.580527
## fac_CS_deliveryFac without CS delivery 0.005160 **
## fac_CS_deliveryHome 0.504683
## antenatal_visits1-4 visits 0.015719 *
## antenatal_visits5+ Visits 0.012862 *
## PostnatalYes 0.065654 .
## child_wanted2Later 0.210351
## child_wanted3Not at all 0.136014
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1380.5 on 3664 degrees of freedom
## Residual deviance: 1258.3 on 3630 degrees of freedom
## AIC: 1279.6
##
## Number of Fisher Scoring iterations: 7exp(cbind(coef(fit), confint(fit)))## 2.5 % 97.5 %
## (Intercept) 0.02080578 0.0001095749 0.3001960
## age_fb20+years 0.61820910 0.4185861395 0.8942905
## child_sexMale 1.93146037 1.3846750199 2.7240020
## b_order3rd or later 1.56300030 0.9241865038 2.8026020
## b_interval> 4 yrs 0.44187182 0.2525963594 0.7702819
## b_interval2-4 yrs 0.43675306 0.2939902563 0.6532502
## time_water 0.99949050 0.9986141820 1.0002749
## water_sourceUnimproved 0.55455604 0.3868734248 0.7880703
## toilet_facilityUnimproved 1.57299305 0.7403141889 3.8394442
## births_5ys 1.20187554 0.9075571929 1.5864558
## residenceUrban 0.81260929 0.3012050863 1.9346011
## educ1Primary 0.78803172 0.5019038991 1.2032709
## educ2Secodary and Higher 1.75313554 0.6408699442 4.2382907
## wealthindex2Middle 0.94189588 0.6065234789 1.4384965
## wealthindex3Rich 1.09683514 0.7199280001 1.6569856
## contraceptive_useNo 1.61984720 1.0930608963 2.4527838
## regionAddis Ababa 0.84009764 0.0754366179 147.2656061
## regionAfar 0.59374321 0.0198139302 121.6787426
## regionAmhara 0.72761968 0.0722655282 125.1408307
## regionBen-Gumuz 1.26271706 0.1306716556 215.2327026
## regionDire Dawa 0.70941720 0.0614443976 125.3148824
## regionGambella 0.76780966 0.0310331079 153.2345799
## regionHarari 1.18917805 0.1209291337 203.4925052
## regionSNNP 0.87382271 0.0005402771 267.0524521
## regionSomali 1.23325278 0.0465655535 253.4901450
## regionTigray 0.92102161 0.0012610136 258.4563288
## bmi_recOverweight 0.66188890 0.2542241914 1.4465225
## bmi_recUnderweight 1.11470246 0.7504986095 1.6243882
## fac_CS_deliveryFac without CS delivery 2.89032825 1.3558753255 6.0674382
## fac_CS_deliveryHome 1.20123224 0.7165662582 2.1158769
## antenatal_visits1-4 visits 0.63407704 0.4360630965 0.9144940
## antenatal_visits5+ Visits 0.45680413 0.2374622195 0.8223529
## PostnatalYes 0.40304688 0.1315677820 0.9516973
## child_wanted2Later 0.73704909 0.4470978512 1.1662373
## child_wanted3Not at all 1.40553881 0.8836990276 2.1695625Fit Cox Proportional Hazard Model for CHILD MORTALITY
## Stratified 1 - level Cluster Sampling design (with replacement)
## With (616) clusters.
## svydesign(ids = ~yrpsu, strata = ~yrstratum, weights = ~wt, data = train.m[is.na(train.m$wt) ==
## F, ], nest = T)## Call:
## svycoxph(formula = Surv(death.age, d.event) ~ age_fb + child_sex +
## b_order + b_interval + time_water + water_source + toilet_facility +
## births_5ys + residence + educ + wealthindex + contraceptive_use +
## region + bmi_rec + fac_CS_delivery + antenatal_visits + Postnatal +
## child_wanted, design = des3)
##
## n= 3665, number of events= 130
##
## coef exp(coef) se(coef)
## age_fb20+years -0.4197088 0.6572382 0.3513775
## child_sexMale 0.6518537 1.9190950 0.2645349
## b_order3rd or later 0.3838926 1.4679877 0.4111089
## b_interval> 4 yrs -0.4687223 0.6258013 0.4721817
## b_interval2-4 yrs -0.6234803 0.5360755 0.3214383
## time_water -0.0004918 0.9995084 0.0004743
## water_sourceUnimproved -0.5415862 0.5818246 0.2974175
## toilet_facilityUnimproved 0.4704848 1.6007701 0.5636677
## births_5ys 0.4421816 1.5560984 0.2644649
## residenceUrban -0.2182127 0.8039544 0.5250125
## educ1Primary -0.2005803 0.8182558 0.3815028
## educ2Secodary and Higher 0.5866950 1.7980361 0.7030814
## wealthindex2Middle -0.0559098 0.9456245 0.3879999
## wealthindex3Rich 0.0915972 1.0959232 0.3258061
## contraceptive_useNo 0.5633544 1.7565547 0.2913205
## regionAddis Ababa -0.1011332 0.9038127 0.5793143
## regionAfar -0.4633316 0.6291839 0.5958243
## regionAmhara -0.2460628 0.7818731 0.6001792
## regionBen-Gumuz 0.3052109 1.3569112 0.6021164
## regionDire Dawa -0.2625471 0.7690901 0.5527513
## regionGambella -0.2497411 0.7790025 0.6076630
## regionHarari 0.2080125 1.2312286 0.5576088
## regionSNNP -0.0983328 0.9063472 0.7021022
## regionSomali 0.1834710 1.2013801 1.0048936
## regionTigray -0.0317239 0.9687740 0.6144651
## bmi_recOverweight -0.4221119 0.6556606 0.6141702
## bmi_recUnderweight 0.0724606 1.0751504 0.3092018
## fac_CS_deliveryFac without CS delivery 1.0245328 2.7857935 0.5649788
## fac_CS_deliveryHome 0.1073052 1.1132740 0.4083747
## antenatal_visits1-4 visits -0.4404657 0.6437365 0.2685589
## antenatal_visits5+ Visits -0.7595519 0.4678760 0.4716453
## PostnatalYes -0.8757972 0.4165298 0.6870792
## child_wanted2Later -0.3361474 0.7145178 0.4795792
## child_wanted3Not at all 0.3416369 1.4072492 0.3268126
## z Pr(>|z|)
## age_fb20+years -1.194 0.2323
## child_sexMale 2.464 0.0137 *
## b_order3rd or later 0.934 0.3504
## b_interval> 4 yrs -0.993 0.3209
## b_interval2-4 yrs -1.940 0.0524 .
## time_water -1.037 0.2999
## water_sourceUnimproved -1.821 0.0686 .
## toilet_facilityUnimproved 0.835 0.4039
## births_5ys 1.672 0.0945 .
## residenceUrban -0.416 0.6777
## educ1Primary -0.526 0.5991
## educ2Secodary and Higher 0.834 0.4040
## wealthindex2Middle -0.144 0.8854
## wealthindex3Rich 0.281 0.7786
## contraceptive_useNo 1.934 0.0531 .
## regionAddis Ababa -0.175 0.8614
## regionAfar -0.778 0.4368
## regionAmhara -0.410 0.6818
## regionBen-Gumuz 0.507 0.6122
## regionDire Dawa -0.475 0.6348
## regionGambella -0.411 0.6811
## regionHarari 0.373 0.7091
## regionSNNP -0.140 0.8886
## regionSomali 0.183 0.8551
## regionTigray -0.052 0.9588
## bmi_recOverweight -0.687 0.4919
## bmi_recUnderweight 0.234 0.8147
## fac_CS_deliveryFac without CS delivery 1.813 0.0698 .
## fac_CS_deliveryHome 0.263 0.7927
## antenatal_visits1-4 visits -1.640 0.1010
## antenatal_visits5+ Visits -1.610 0.1073
## PostnatalYes -1.275 0.2024
## child_wanted2Later -0.701 0.4834
## child_wanted3Not at all 1.045 0.2959
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## exp(coef) exp(-coef) lower .95
## age_fb20+years 0.6572 1.5215 0.3301
## child_sexMale 1.9191 0.5211 1.1427
## b_order3rd or later 1.4680 0.6812 0.6558
## b_interval> 4 yrs 0.6258 1.5980 0.2480
## b_interval2-4 yrs 0.5361 1.8654 0.2855
## time_water 0.9995 1.0005 0.9986
## water_sourceUnimproved 0.5818 1.7187 0.3248
## toilet_facilityUnimproved 1.6008 0.6247 0.5303
## births_5ys 1.5561 0.6426 0.9267
## residenceUrban 0.8040 1.2439 0.2873
## educ1Primary 0.8183 1.2221 0.3874
## educ2Secodary and Higher 1.7980 0.5562 0.4532
## wealthindex2Middle 0.9456 1.0575 0.4420
## wealthindex3Rich 1.0959 0.9125 0.5787
## contraceptive_useNo 1.7566 0.5693 0.9924
## regionAddis Ababa 0.9038 1.1064 0.2904
## regionAfar 0.6292 1.5894 0.1957
## regionAmhara 0.7819 1.2790 0.2411
## regionBen-Gumuz 1.3569 0.7370 0.4169
## regionDire Dawa 0.7691 1.3002 0.2603
## regionGambella 0.7790 1.2837 0.2368
## regionHarari 1.2312 0.8122 0.4128
## regionSNNP 0.9063 1.1033 0.2289
## regionSomali 1.2014 0.8324 0.1676
## regionTigray 0.9688 1.0322 0.2905
## bmi_recOverweight 0.6557 1.5252 0.1967
## bmi_recUnderweight 1.0752 0.9301 0.5865
## fac_CS_deliveryFac without CS delivery 2.7858 0.3590 0.9205
## fac_CS_deliveryHome 1.1133 0.8983 0.5000
## antenatal_visits1-4 visits 0.6437 1.5534 0.3803
## antenatal_visits5+ Visits 0.4679 2.1373 0.1856
## PostnatalYes 0.4165 2.4008 0.1083
## child_wanted2Later 0.7145 1.3995 0.2791
## child_wanted3Not at all 1.4072 0.7106 0.7416
## upper .95
## age_fb20+years 1.309
## child_sexMale 3.223
## b_order3rd or later 3.286
## b_interval> 4 yrs 1.579
## b_interval2-4 yrs 1.007
## time_water 1.000
## water_sourceUnimproved 1.042
## toilet_facilityUnimproved 4.832
## births_5ys 2.613
## residenceUrban 2.250
## educ1Primary 1.728
## educ2Secodary and Higher 7.133
## wealthindex2Middle 2.023
## wealthindex3Rich 2.075
## contraceptive_useNo 3.109
## regionAddis Ababa 2.813
## regionAfar 2.023
## regionAmhara 2.535
## regionBen-Gumuz 4.416
## regionDire Dawa 2.272
## regionGambella 2.563
## regionHarari 3.673
## regionSNNP 3.589
## regionSomali 8.611
## regionTigray 3.230
## bmi_recOverweight 2.185
## bmi_recUnderweight 1.971
## fac_CS_deliveryFac without CS delivery 8.431
## fac_CS_deliveryHome 2.479
## antenatal_visits1-4 visits 1.090
## antenatal_visits5+ Visits 1.179
## PostnatalYes 1.601
## child_wanted2Later 1.829
## child_wanted3Not at all 2.670
##
## Concordance= 0.737 (se = 0.032 )
## Rsquare= NA (max possible= NA )
## Likelihood ratio test= NA on 34 df, p=NA
## Wald test = 123.8 on 34 df, p=4e-12
## Score (logrank) test = NA on 34 df, p=NA## [,1]
## sensitivity 0.57142857
## specificity 0.72992701
## cutoff 0.04504857## Confusion Matrix and Statistics
##
##
## 0 1
## 0 1214 293
## 1 36 27
##
## Accuracy : 0.7904
## 95% CI : (0.7695, 0.8103)
## No Information Rate : 0.7962
## P-Value [Acc > NIR] : 0.7255
##
## Kappa : 0.0793
## Mcnemar's Test P-Value : <2e-16
##
## Sensitivity : 0.97120
## Specificity : 0.08438
## Pos Pred Value : 0.80557
## Neg Pred Value : 0.42857
## Prevalence : 0.79618
## Detection Rate : 0.77325
## Detection Prevalence : 0.95987
## Balanced Accuracy : 0.52779
##
## 'Positive' Class : 0
## ## [,1]
## sensitivity 0.5873016
## specificity 0.6589250
## cutoff 1.2832007## Confusion Matrix and Statistics
##
##
## 0 1
## 0 790 717
## 1 21 42
##
## Accuracy : 0.5299
## 95% CI : (0.5049, 0.5549)
## No Information Rate : 0.5166
## P-Value [Acc > NIR] : 0.1503
##
## Kappa : 0.0303
## Mcnemar's Test P-Value : <2e-16
##
## Sensitivity : 0.97411
## Specificity : 0.05534
## Pos Pred Value : 0.52422
## Neg Pred Value : 0.66667
## Prevalence : 0.51656
## Detection Rate : 0.50318
## Detection Prevalence : 0.95987
## Balanced Accuracy : 0.51472
##
## 'Positive' Class : 0
## Random Forest Model - Under-5 Mortality
##
## Call:
## randomForest(formula = chmort ~ v201 + v445 + v024 + v025 + v106 + v445 + v012 + v013 + v208 + v137 + bord + b11 + v190 + v201 + v113 + v115 + v116 + v364 + v212 + b4 + v106 + m15 + v130 + m1 + m10 + m14 + m15 + m70, data = train.m)
## Type of random forest: classification
## Number of trees: 500
## No. of variables tried at each split: 4
##
## OOB estimate of error rate: 3.22%
## Confusion matrix:
## 0 1 class.error
## 0 3527 8 0.002263083
## 1 110 20 0.846153846Performance of Random Forest Model on Testing Data
pred = predict(rf.m, newdata=test.m)
accuracy <- table(pred, test.m[,"chmort"])
sum(diag(accuracy))/sum(accuracy)## [1] 0.9636943confusionMatrix(data=pred, as.factor(test.m$chmort))## Confusion Matrix and Statistics
##
## Reference
## Prediction 0 1
## 0 1503 53
## 1 4 10
##
## Accuracy : 0.9637
## 95% CI : (0.9532, 0.9724)
## No Information Rate : 0.9599
## P-Value [Acc > NIR] : 0.2428
##
## Kappa : 0.2488
## Mcnemar's Test P-Value : 2.047e-10
##
## Sensitivity : 0.9973
## Specificity : 0.1587
## Pos Pred Value : 0.9659
## Neg Pred Value : 0.7143
## Prevalence : 0.9599
## Detection Rate : 0.9573
## Detection Prevalence : 0.9911
## Balanced Accuracy : 0.5780
##
## 'Positive' Class : 0
## test_prob = predict(fit, newdata = test.m, type = "response")
test_roc = roc(test.m$chmort ~ test_prob, plot = TRUE, print.auc = TRUE,percent=TRUE)
test_prob = predict(cox.child, newdata = test.m, type = "risk")
test_roc2 = roc(test.m$chmort ~ test_prob, plot = TRUE, print.auc = TRUE,percent=TRUE)
test_prob5 = predict(rf.m, newdata=test.m, type = "prob")
test_roc5= roc(test.m$chmort, test_prob5[,c(2)],plot = TRUE, col='darkgoldenrod4', print.auc = TRUE,percent=TRUE)
plot(test_roc, print.auc = TRUE, col='blue',percent=TRUE, print.auc.y = 41, print.auc.x = 0)
plot(test_roc2 , add = TRUE, col='red', print.auc = TRUE, print.auc.y = 34, print.auc.x = 0)
plot(test_roc5 , add = TRUE, col='darkgoldenrod4', print.auc = TRUE, print.auc.y = 27, print.auc.x = 0)
abline(a=0, b=1, lty=2, lwd=3, col="black")
legend("topleft", legend = c("Logistic-Regression", "Cox-Proportional Hazard",
"Random Forest-Model") , pch = 15, bty = 'n', col = c("blue","red", "darkgoldenrod4"))