library(tidyverse)
## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr     1.2.1     ✔ readr     2.2.0
## ✔ forcats   1.0.1     ✔ stringr   1.6.0
## ✔ ggplot2   4.0.3     ✔ tibble    3.3.1
## ✔ lubridate 1.9.5     ✔ tidyr     1.3.2
## ✔ purrr     1.2.2     
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag()    masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
library(readxl)
library(ggplot2)
library(lmtest)
## Loading required package: zoo
## 
## Attaching package: 'zoo'
## 
## The following objects are masked from 'package:base':
## 
##     as.Date, as.Date.numeric
library(car)
## Loading required package: carData
## 
## Attaching package: 'car'
## 
## The following object is masked from 'package:dplyr':
## 
##     recode
## 
## The following object is masked from 'package:purrr':
## 
##     some
FinalData<-read_excel("FinalData.xlsx")
FinalData1<-FinalData |> drop_na(`PCP`) %>% filter(`PCP`>0)
FinalData2<-FinalData1 |> drop_na('MHP') %>% filter(`MHP`>0)
FinalData3<-FinalData2 |> drop_na('DEN') %>% filter(`DEN`>0)
FinalData4<-FinalData3 |> drop_na('PHR') %>% filter(`PHR`>0)
FinalData5<-FinalData4 |> drop_na('UIN') %>% filter(`UIN`>0)
FinalData6<-FinalData5 |> drop_na('SHCB') %>% filter(`SHCB`>0)
FinalData7<-FinalData6 |> drop_na('UNEM') %>% filter(`UNEM`>0)
FinalData8<-FinalData7 |> drop_na('DOD') %>% filter(`DOD`>0)
FinalData9<-FinalData8 |> drop_na('UNIN') %>% filter(`UNIN`>0)
CountyFinal <- FinalData9 %>% dplyr::select(`DOD`, `SHCB`, `MHP`,`UNIN`) %>% arrange(-`DOD`, `SHCB`, `MHP`,`UNIN`)

#Correlation: Model H

cor(CountyFinal)
##            DOD      SHCB       MHP      UNIN
## DOD  1.0000000 0.3157972 0.9545255 0.9767553
## SHCB 0.3157972 1.0000000 0.3504348 0.3390718
## MHP  0.9545255 0.3504348 1.0000000 0.9224518
## UNIN 0.9767553 0.3390718 0.9224518 1.0000000
County_model <-lm(DOD~SHCB+MHP+UNIN,data=CountyFinal)

#Summary of Data: Model G

summary(County_model)
## 
## Call:
## lm(formula = DOD ~ SHCB + MHP + UNIN, data = CountyFinal)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -426.32   -5.14    4.21   16.72  223.51 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept) 44.1453207 26.2084072   1.684   0.0951 .  
## SHCB        -4.6372428  2.3631425  -1.962   0.0524 .  
## MHP          0.1049338  0.0117132   8.959 1.42e-14 ***
## UNIN         0.0025055  0.0001576  15.903  < 2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 59.54 on 104 degrees of freedom
## Multiple R-squared:  0.9742, Adjusted R-squared:  0.9735 
## F-statistic:  1310 on 3 and 104 DF,  p-value: < 2.2e-16

#Linearity: Model A

plot(County_model,which=1)

#The model is not linear based on visual inspection.

#Linear Transformation: Model A-1

County_model_log3<-lm(DOD~UNIN+SHCB+MHP,data=CountyFinal)
plot(County_model_log3,which=1)

#Model was transformed but final result is still nonlinear per visual inspection.

#Independence of Errors: Model B

durbinWatsonTest(County_model)
##  lag Autocorrelation D-W Statistic p-value
##    1     -0.08905871      2.040188   0.896
##  Alternative hypothesis: rho != 0

#The models errors are independent as the p-value is .966 which is greater than the standard 0.05.

#Homoscedasticity: Model C

plot(County_model,which=3)

#This model is not homoscedastic per visual inspection.

#Homoscedascity Transformation: Model C-1

County_model_log<-lm(log(DOD)~MHP+SHCB+UNIN,data=CountyFinal)
plot(County_model_log,which=3)

#Model is mildly improved for homoscedasticity per visual inspection.

#Homoscedasticity: Model D

bptest(County_model)
## 
##  studentized Breusch-Pagan test
## 
## data:  County_model
## BP = 32.472, df = 3, p-value = 4.163e-07

#Model is not homoscedastic per visual inspection

#Homoscedasticity Transformation: Model D-1

bptest(County_model_log)
## 
##  studentized Breusch-Pagan test
## 
## data:  County_model_log
## BP = 13.009, df = 3, p-value = 0.004618

#Model is not homoscedastic per visual inspection

#Normality of Residuals: Model E

plot(County_model,which=2)

#Model is somewhat abnormal by visual inspection as the data points stray from the line.

#Normality of Residuals Transformation: Model E-1

County_model <-lm(DOD~SHCB+MHP+UNIN,data=CountyFinal)
shapiro.test(County_model$residuals)
## 
##  Shapiro-Wilk normality test
## 
## data:  County_model$residuals
## W = 0.61897, p-value = 2.515e-15
County_model_log2<-lm(log(DOD)~MHP,SHCB,UNIN,data=CountyFinal)
shapiro.test(County_model_log2$residuals)
## 
##  Shapiro-Wilk normality test
## 
## data:  County_model_log2$residuals
## W = 0.89023, p-value = 2.165e-07
plot(County_model_log2,which=2)

#Model is still abnormal by visual inspection.

#Multicolinearity: Model F

vif(County_model)
##     SHCB      MHP     UNIN 
## 1.142181 6.780560 6.720534

#Model does not have severe multivolinearity as results are less than the threshhold of 10.

#Multicolinearity: Model F-1

kitchen_sink<-lm(DOD~UNIN+SHCB+MHP,data=CountyFinal)
vif(kitchen_sink)
##     UNIN     SHCB      MHP 
## 6.720534 1.142181 6.780560