datacleaning_descriptivestats_CIND820.Rmd

---
title: Data cleaning and descriptive statistics for CIND820
output: file:///E:/Data_Analytics/Final%20Project/BRFSS/BRFSS_2015/datacleaning_descriptivestats_CIND820.nb.html
--- 

```{r}
library(foreign)
```

```{r}
brfss_data <- read.xport("C:/Users/mmkea/Documents/2015_raw_data.xpt")
```

```{r}
head(brfss_data)
```


First, limit the data to our age groups of interest (eg, 60 years and older). 
```{r}
unique(brfss_data$X_AGEG5YR) #9-13 is 60-80+ ; 14 is refused or don't know
brfss_data_60YR <- subset(brfss_data, brfss_data$X_AGEG5YR > 8 & brfss_data$X_AGEG5YR < 14)
head(brfss_data_60YR) #202412 individuals
```
Next, let's reduce our data-set down from 330 variables to 45 variables. See data dictionary for full explanation. 
```{r}
data <- subset(brfss_data_60YR, 
               select=c(SEX, MARITAL,NUMADULT, HHADULT, CHILDREN,
                        X_RACE,
             X_AGEG5YR, X_BMI5, X_EDUCAG, EMPLOY1, INCOME2, RENTHOM1, VETERAN3, X_RFHLTH,
             PHYSHLTH, MENTHLTH, HLTHPLN1, MEDCOST, CHECKUP1, X_RFHYPE5,
             X_RFCHOL, X_MICHD, CVDSTRK3, X_ASTHMS1, CHCSCNCR, CHCOCNCR,
             CHCCOPD1, X_DRDXAR1, ADDEPEV2, CHCKIDNY, DIABETE3, X_SMOKER3,
             QLACTLM2, USEEQUIP, BLIND, DECIDE, ALCDAY5, AVEDRNK2, 
              X_FRUTSUM, BEANDAY_, X_VEGESUM, PA1MIN_, EXRACT11,
             STRENGTH, PA1VIGM_, FLUSHOT6))

```

Reducing the HHADULT and NUMADULT variables to one ADULT variable. 
```{r}
unique(data$HHADULT) 
sum(is.na(data$HHADULT))
unique(data$NUMADULT) 
sum(is.na(data$NUMADULT))
table(data$NUMADULT)

#Replacing the NA from HHADULT where there are values in NUMADULT 
#and combining into a new variable. 
data$ADULT <- data$HHADULT
data$ADULT[!is.na(data$NUMADULT)] <-  data$NUMADULT[!is.na(data$NUMADULT)]
table(data$ADULT)

#Now, let's remove the "77" and "99" responses in ADULT 
#These are our "refused" and "don't know" codes. 
#data <- subset(data, data$ADULT<77 | is.na(data$ADULT))
data$ADULT[data$ADULT==77] <- NA
data$ADULT[data$ADULT==99] <- NA

#And finally, let's remove the HHADULT and NUMADULT columns. 
data <- subset(data, select= -c(NUMADULT, HHADULT))
head(data)

```
Next, let's condense our cancer variables CHCSCNCR ("Ever had skin cancer") and CHCOCNCR ("Ever had other cancer") to CNCER ("Ever had cancer"). 
```{r}
#Checking the unique numbers. 
unique(data$CHCSCNCR) #1,2,7,9,NA
unique(data$CHCOCNCR) #1,2,7,9,NA

#Creating the new variables. 
data$CNCER <- NA
data$CNCER[data$CHCSCNCR==1 | data$CHCOCNCR==1] <- 1
data$CNCER[data$CHCSCNCR==2 & data$CHCOCNCR==2] <- 2
table(data$CNCER)

data <- subset(data, select=-c(CHCSCNCR, CHCOCNCR))
```
Calculating average alcohol consumption per week.  
```{r}
#ALCDAY5; 101-199 is 1__ days per week; 201-299 is 2__ days per month. #AVEDRNK2 1-76 (average consumption when drinking).
unique(data$ALCDAY5)
boxplot(data$AVEDRNK2)

#First, replace 888 with 0 and remove 777/999 responses.
data$ALCDAY5[data$ALCDAY5==888] <- 0
data$ALCDAY5[data$ALCDAY5==777] <- NA
data$ALCDAY5[data$ALCDAY5==999] <- NA

data$AVEDRNK2[data$AVEDRNK2==77] <- NA
data$AVEDRNK2[data$AVEDRNK2==99] <- NA
# data <- subset(data, data$ALCDAY5<777 | is.na(data$ALCDAY5))
# data <- subset(data, data$AVEDRNK2<77 | is.na(data$AVEDRNK2))

#Create a new column. 
data$AVGDRNK <- NA

#Creating a function to calculate average alcohol consumption 
#per week. 
average_drink<-function(ALCDAY5, AVEDRNK2){
  ifelse(ALCDAY5==0, 0, ifelse(ALCDAY5>100 & ALCDAY5<108, (ALCDAY5 - 100)*AVEDRNK2, ifelse(ALCDAY5>200 & ALCDAY5<232, ((ALCDAY5 - 200)/4.2857)*AVEDRNK2, NA)))
}

data$AVGDRNK <- with(data, average_drink(data$ALCDAY5, data$AVEDRNK2))
boxplot(data$AVGDRNK) #may need to remove those top two outliers.

#Removing the original variables. 
data <- subset(data, select=-c(ALCDAY5, AVEDRNK2))
```
Let's also use the same formulas above to recode STRENGTH to frequency per week. 

```{r}
#STRENGTH; 101-199 is 1__ times per week; 201-299 is 2__ times per 
#month for strength training frequency.
unique(data$STRENGTH)
data$STRENGTH[data$STRENGTH==888] <- 0
data$STRENGTH[data$STRENGTH==777] <- NA
data$STRENGTH[data$STRENGTH==999] <- NA
#data <- subset(data, data$STRENGTH < 777 | is.na(data$STRENGTH))

data$STRNGTH <- NA

#Creating a function to calclulate strength training frequency in 
#a week. TK UPDATE TO REMOVE 888. 
avg_strength_training<-function(STRENGTH){
  ifelse(STRENGTH==0, 0, ifelse(STRENGTH>100 & STRENGTH<200, STRENGTH-100, ifelse(STRENGTH>200 & STRENGTH<300, (STRENGTH-200)/4.2857, NA)))
}

data$STRNGTH <- with(data, avg_strength_training(data$STRENGTH))

data <- subset(data, select=-c(STRENGTH))
unique(data$STRNGTH)
```
Now, let's recode some of our "88" responses (eg, "none") responses to 0. We'll also remove the 7/9 or 77/99 responses here as well. 
```{r}
#CHILDREN. 
unique(data$CHILDREN) 
data$CHILDREN[data$CHILDREN==88] <- 0
data$CHILDREN[data$CHILDREN==99] <- NA
unique(data$CHILDREN)

#PHYSHLTH. 
unique(data$PHYSHLTH) 
data$PHYSHLTH[data$PHYSHLTH==88] <- 0
data$PHYSHLTH[data$PHYSHLTH==77 | data$PHYSHLTH==99] <- NA
unique(data$PHYSHLTH)

#MENHLTH. 
unique(data$MENTHLTH) 
data$MENTHLTH[data$MENTHLTH==88] <- 0
data$MENTHLTH[data$MENTHLTH==77 | data$MENTHLTH==99] <- NA
unique(data$MENTHLTH)

```

And finally, let's convert all responses with "refused" or "don't know" to NA. 
```{r}
#SEX. 
unique(data$SEX) #nothing to remove

#MARITAL. 
unique(data$MARITAL)
data$MARITAL[data$MARITAL==9] <- NA

#X_Race.
unique(data$X_RACE)
data$X_RACE[data$X_RACE==9] <- NA

#X_EDUCAG. 
unique(data$X_EDUCAG)
data$X_EDUCAG[data$X_EDUCAG==9] <- NA

#EMPLOY1. 
unique(data$EMPLOY1)
data$EMPLOY1[data$EMPLOY1==9] <- NA

#INCOME2.
unique(data$INCOME2)
data$INCOME2[data$INCOME2==77 | data$INCOME2==99] <- NA

#RENTHOM1 
unique(data$RENTHOM1)
data$RENTHOM1[data$RENTHOM1==7 | data$RENTHOM1==9] <- NA

#VETERAN3 
unique(data$VETERAN3)
data$VETERAN3[data$VETERAN3==7 | data$VETERAN3==9] <- NA

#HLTHPLN1. 
unique(data$HLTHPLN1)
data$HLTHPLN1[data$HLTHPLN1==9 | data$HLTHPLN1==7] <- NA

#MEDCOST
unique(data$MEDCOST)
data$MEDCOST[data$MEDCOST==7 | data$MEDCOST==9] <- NA

#CHECKUP1. 
unique(data$CHECKUP1)
data$CHECKUP1[data$CHECKUP1==8] <- 0
data$CHECKUP1[data$CHECKUP1==7 | data$CHECKUP1==9] <- NA
unique(data$CHECKUP1)

#X _RFHYPE5. 
unique(data$X_RFHYPE5)
data$X_RFHYPE5[data$X_RFHYPE5==9] <- NA

#X _RFCHOL. 
unique(data$X_RFCHOL)
data$X_RFCHOL[data$X_RFCHOL==9] <- NA

#X_MICHD.
unique(data$X_MICHD) #none to remove 

#CVDSTRK3. 
unique(data$CVDSTRK3)
data$CVDSTRK3[data$CVDSTRK3==7 | data$CVDSTRK3==9] <- NA

#X_ASTHMS1. 
unique(data$X_ASTHMS1)
data$X_ASTHMS1[data$X_ASTHMS1==9] <- NA

#CHCCOPD1. 
unique(data$CHCCOPD1)
data$CHCCOPD1[data$CHCCOPD1==7 | data$CHCCOPD1==9] <- NA

#ADDEPEV2. 
unique(data$ADDEPEV2)
data$ADDEPEV2[data$ADDEPEV2==7 | data$ADDEPEV2==9] <- NA

#CHCKIDNY. 
unique(data$CHCKIDNY)
data$CHCKIDNY[data$CHCKIDNY==7 | data$CHCKIDNY==9] <- NA

#DIABETE3. 
unique(data$DIABETE3)
data$DIABETE3[data$DIABETE3==7 | data$DIABETE3==9] <- NA

#X_SMOKER3. 
unique(data$X_SMOKER3)
data$X_SMOKER3[data$X_SMOKER3==9] <- NA

#QLACTLM2. 
unique(data$QLACTLM2)
data$QLACTLM2[data$QLACTLM2==7 | data$QLACTLM2==9] <- NA

#USEEQUIP. 
unique(data$USEEQUIP)
data$USEEQUIP[data$USEEQUIP==7 | data$USEEQUIP==9] <- NA

#BLIND. 
unique(data$BLIND)
data$BLIND[data$BLIND==7 | data$BLIND==9] <- NA

#DECIDE. 
unique(data$DECIDE)
data$DECIDE[data$DECIDE==7 | data$DECIDE==9] <- NA

#EXRACT11. 
unique(data$EXRACT11)
data$EXRACT11[data$EXRACT11==77 | data$EXRACT11==99] <- NA
unique(data$EXRACT11)

#FLUSHOT6. 
unique(data$FLUSHOT6)
data$FLUSHOT6[data$FLUSHOT6==7 | data$FLUSHOT6==9] <- NA

```
Now, for our target attribute (X_RFHLTH) let's remove BOTH the 'NA' and '9' values. 
```{r}
#X_RFHLTH. Subset function removes all NA in our target attribute. 
table(data$X_RFHLTH, useNA="ifany")
data <- subset(data, data$X_RFHLTH<9)
table(data$X_RFHLTH, useNA="ifany")
```
Let's take a look at our proportions of NA values within each column of the dataset. 
```{r}
lapply(data, function(var) {(sum(is.na(var))/length(var))*100}) 
#INCOME and EXRACT11 can be recoded since they are categorical. We'll 
#leave PA1MIN_ and PA1VIGM_ until we have a better understanding of
#the relationship with X_RFHLTH. 
```

```{r}
#Recoding the NA values in INCOME and EXTRACT.
data$INCOME2[is.na(data$INCOME2)]<-0
data$EXRACT11[is.na(data$EXRACT11)]<-0
```

```{r}
#Number of respondents with 50% NA values (31 missing columns).
sum(rowSums(is.na(data))>=24) # 9

#Number of respondents with 30% or more NA values (19 columns). 
sum(rowSums(is.na(data))>=14) # 4366

#Number of respondents with 15% or more NA values (6 columns)
sum(rowSums(is.na(data))>5) # 15265

data_clean<-data[rowSums(is.na(data))<5,] #data with less than 15%

#Percent of missing values in each column.  
lapply(data_clean, function(var) {(sum(is.na(var))/length(var))*100}) 
```

OK. Let's clean our numeric variables. Our numeric attributes include CHILDREN, ADULT, X_BMI5, PHYSHLTH, MENHLTH, AVGDRNK, X_FRUTSUM, BEANDAY_, X_VEGESUM, PA1MIN, STRENGTH, PA1VIGM_. The rule of thumb is that outliers in the 99th percentiles (eg, top and bottom 1% of the data) will be brought into the 99th percentile. Exceptions will be allowed if the value of the 99th percentile seems plausible (eg, plausible to eat or drink a certain amount).  

```{r}
str(data_clean) #BMI, FRUT, VEG, BEAN variables have incorrect values
data_clean$X_BMI5 <- data_clean$X_BMI5/100
data_clean$X_FRUTSUM <- data_clean$X_FRUTSUM/100
data_clean$X_VEGESUM <- data_clean$X_VEGESUM/100
data_clean$BEANDAY_ <- data_clean$BEANDAY_/100
```
CHILDREN. 
```{r}
#CHILDREN. 
summary(data_clean$CHILDREN) #noticeable skew with outliers

#Winsorizing all outliers. 
quantile(data_clean$CHILDREN, .999, na.rm=TRUE) #99th percentile #seemed low, so we're using the .1th percentile  
data_clean$CHILDREN[data_clean$CHILDREN>4]<-4
unique(data_clean$CHILDREN)

#Assessing cleaned data. 
summary(data_clean$CHILDREN) 
sd(data_clean$CHILDREN, na.rm = TRUE)
library(modeest) #mode
mfv(data_clean$CHILDREN)
hist(data_clean$CHILDREN,
     col = "darksalmon",
     xlim=c(0,4),
     ylim=c(0,140000),
     main = "Number of Children In House",
     xlab="Number of Children")
```

ADULT. 
```{r}
#ADULT. 
summary(data_clean$ADULT)

#Converting outliers to "4" which from now on will mean 4 or more 
#adults. 
quantile(data_clean$ADULT, .99, na.rm=TRUE) #4 
data_clean$ADULT[data_clean$ADULT>4]<-4
unique(data_clean$ADULT)

#Assessing cleaned data. 
summary(data_clean$ADULT) #Mean = 1.7; Median = 2.0
sd(data_clean$ADULT, na.rm = TRUE) #SD = 
hist(data_clean$ADULT,
     col = "darksalmon",
     xlim=c(1,4),
     ylim=c(0,80000),
     xlab="Number of Adults (Including Respondent)")
```
BMI. 
```{r}
#BMI. 
summary(data_clean$X_BMI5)

#Winsorizing. 
quantile(data_clean$X_BMI5, c(.01, .99), na.rm=TRUE)
data_clean$X_BMI5[data_clean$X_BMI5<18.02]<-17.92
data_clean$X_BMI5[data_clean$X_BMI5>44.08]<-46.48

#Assessing distribution.
mfv(data_clean$X_BMI5, na_rm = TRUE)
sd(data_clean$X_BMI5, na.rm = TRUE)
boxplot(data_clean$X_BMI5,
     col = "darksalmon",
     ylab="Calculated BMI") 
```
PHYSHLTH. 
```{r}
summary(data_clean$PHYSHLTH)
quantile(data_clean$PHYSHLTH, .99, na.rm=TRUE) #left as is
mfv(data_clean$PHYSHLTH, na_rm = TRUE)
sd(data_clean$PHYSHLTH, na.rm = TRUE)
boxplot(data_clean$PHYSHLTH,
     col = "darksalmon",
     ylab="Number of Days") #may switch to violin plot
```
MENHLTH. 
```{r}
summary(data_clean$MENTHLTH)
quantile(data_clean$MENTHLTH, .99, na.rm=TRUE) #left as is 
mfv(data_clean$MENTHLTH, na_rm = TRUE)
sd(data_clean$MENTHLTH, na.rm = TRUE)
boxplot(data_clean$MENTHLTH,
     col = "darksalmon",
     ylab="Number of Days") #may switch to violin plot
```

AVGDRINK. 
```{r}
summary(data_clean$AVGDRNK)
quantile(data_clean$AVGDRNK, .99, na.rm=TRUE)  

summary(data_clean$AVGDRNK)
mfv(data_clean$AVGDRNK)
sd(data_clean$AVGDRNK, na.rm = TRUE)
data_clean$AVGDRNK[data_clean$AVGDRNK>23.33]<-23.33
boxplot(data_clean$AVGDRNK,
     col = "darksalmon",
     ylab="Average Alcohol Consumption")
```
FRUIT. 
```{r}
summary(data_clean$X_FRUTSUM) 
quantile(data_clean$X_FRUTSUM, .999, na.rm=TRUE) #7 seems low; using # the 0.1th percentile 
data_clean$X_FRUTSUM[data_clean$X_FRUTSUM>8]<-8


#Descriptives. 
summary(data_clean$X_FRUTSUM)
mfv(data_clean$X_FRUTSUM)
sd(data_clean$X_FRUTSUM, na.rm = TRUE)
boxplot(data_clean$X_FRUTSUM,
     col = "darksalmon",
     ylab = "Fruit Consumption per Day")

```
VEG. 
```{r}
summary(data_clean$X_VEGESUM) 
quantile(data_clean$X_VEGESUM, .99, na.rm=TRUE) #seems low, using
#the 0.1th percentile 
data_clean$X_VEGESUM[data_clean$X_VEGESUM>11]<-11

#Descriptives. 
summary(data_clean$X_VEGESUM)
mfv(data_clean$X_VEGESUM, na_rm = TRUE)
sd(data_clean$X_VEGESUM, na.rm = TRUE)
boxplot(data_clean$X_VEGESUM,
     col = "darksalmon",
     ylab="Number of Vegetables per Day")
```
BEAN. 
```{r}
summary(data_clean$BEANDAY_) 
quantile(data_clean$BEANDAY_, .999, na.rm=TRUE) #seems low; using .1th #percentile
data_clean$BEANDAY_[data_clean$BEANDAY_>4]<-4

#Descriptives. 
summary(data_clean$BEANDAY_)
mfv(data_clean$BEANDAY_, na_rm = TRUE)
sd(data_clean$BEANDAY_, na.rm = TRUE)
boxplot(data_clean$BEANDAY_,
     col = "darksalmon",
     ylab="Number of Legumes per Day")
```
PA.
```{r}
data_clean$PA1MIN_ <- data_clean$PA1MIN_/7 #change to per day

summary(data_clean$PA1MIN_) #values seem OK
quantile(data_clean$PA1MIN_, .99, na.rm=TRUE)
data_clean$PA1MIN_[data_clean$PA1MIN_>600]<-600

#Descriptives. 
summary(data_clean$PA1MIN_) #values seem OK
mfv(data_clean$PA1MIN_)
sd(data_clean$PA1MIN_, na.rm = TRUE)
boxplot(data_clean$PA1MIN_,
     col = "darksalmon",
     ylab="Minutes of Activity Per Day")
```
STRGNTH. 
```{r}
summary(data_clean$STRNGTH) #values seem OK

#Recoding to "yes" and "no" of 1 and 2. 
table(data_clean$STRNGTH)
data_clean$STRNGTH[data_clean$STRNGTH>0]<-1
data_clean$STRNGTH[data_clean$STRNGTH==0]<-2
table(data_clean$STRNGTH)

table(data_clean$STRNGTH, useNA = "ifany")
mfv(data_clean$STRNGTH)
barplot(table(data_clean$STRNGTH, useNA = "ifany"),
     col = "darksalmon",
     ylab="Frequency",
     names.arg=c("Yes", "No", "Missing")) 
```
PA1VIGM.
```{r}
summary(data_clean$PA1VIGM_) #values seem OK
data_clean$PA1VIGM_ <- data_clean$PA1VIGM_/7 #to per day
quantile(data_clean$PA1VIGM_, .99, na.rm=TRUE)
data_clean$PA1VIGM_[data_clean$PA1VIGM_>274.28]<-274.28

#Descriptives. 
summary(data_clean$PA1VIGM_) #values seem OK
mfv(data_clean$PA1VIGM_)
sd(data_clean$PA1VIGM_, na.rm = TRUE)
boxplot(data_clean$PA1VIGM_,
     col = "darksalmon",
     ylab="Minutes of Vigorous Activity Per Day")
```

Now, Let's run through some descriptives of our categorical variables.
```{r}
str(data_clean)
```
SEX. 
```{r}
#descriptives. 
table(data_clean$SEX, useNA = "ifany")
mfv(data_clean$SEX)

#visual
barplot(table(data_clean$SEX, 
              useNA = "ifany"),
col="darksalmon",
        names.arg= c("Male", "Female"),
        ylab = "Frequency",
        ylim=c(0,80000))
```
MARITAL. 
```{r}
#MARITAL. 
table(data_clean$MARITAL, useNA = "ifany")
mfv(data_clean$MARITAL)
barplot(table(data_clean$MARITAL, 
              useNA = "ifany"),
col="darksalmon",
las=3,
        names.arg= c("Married", "Divorced", "Widowed", "Separated",
                      "Never\nMarried", "Common\nLaw", "Missing"),
        ylab = "Frequency",
        ylim=c(0,80000))
```
RACE. 
```{r}
table(data_clean$X_RACE, useNA = "ifany")
mfv(data_clean$X_RACE)
barplot(table(data_clean$X_RACE, 
              useNA = "ifany"),
col="darksalmon",
        names.arg= c("White", "Black", "Native", "Asian", "Native\nIslander","Other","Multiracial", "Hispanic", "Missing"),
las=3,
        ylab = "Frequency",
        ylim=c(0,120000))
```
AGE. 
```{r}
table(data_clean$X_AGEG5YR, useNA = "ifany")
mfv(data_clean$X_AGEG5YR)
barplot(table(data_clean$X_AGEG5YR, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("60-64", "65-69", "70-74","75-79", "80+"),
        ylab = "Frequency",
        ylim=c(0,40000))
```
EDUCATION. 
```{r}
table(data_clean$X_EDUCAG, useNA = "ifany")
mfv(data_clean$X_EDUCAG)
barplot(table(data_clean$X_EDUCAG, 
              useNA = "ifany"),
        col="darksalmon",
        las=3,
        names.arg=c("No HS\nDiploma", "HS\nDiploma","Attended\nCollege", "College\nDiploma", "Missing"),
        ylab = "Frequency",
        ylim=c(0,60000))
```
EMPLOYMENT. 
```{r}
table(data_clean$EMPLOY1, useNA = "ifany")

#Need to condense some of these categories. 
data_clean$EMPLOY1[data_clean$EMPLOY1==1 | data_clean$EMPLOY1==2] <- 1
data_clean$EMPLOY1[data_clean$EMPLOY1==3]<- 2 
data_clean$EMPLOY1[data_clean$EMPLOY1==4]<- 2
data_clean$EMPLOY1[data_clean$EMPLOY1==8]<- 2
data_clean$EMPLOY1[data_clean$EMPLOY1==7]<-3
data_clean$EMPLOY1[data_clean$EMPLOY1==5 | data_clean$EMPLOY1==6]<- 4

#Descriptives. 
table(data_clean$EMPLOY1, useNA = "ifany")
mfv(data_clean$EMPLOY1)

#Visual. 
barplot(table(data_clean$EMPLOY1, 
              useNA = "ifany"),
        col="darksalmon",
        las=3,
        names.arg= c("Employed", "Not\nEmployed", "Retired","Other", "Missing"),
        ylab = "Frequency",
        ylim=c(0,80000))
```
INCOME. 
```{r}
table(data_clean$INCOME2, useNA = "ifany")
mfv(data_clean$INCOME2)

barplot(table(data_clean$INCOME2, 
              useNA = "ifany"),
        las=3,
        col="darksalmon",
        names.arg= c( "Not\nReported", "<10,000", "10-15,000", "15-20,000","20-25,000", "25-35,000", "35-50,000","50-75,000", ">75,000"),
        ylab = "Frequency",
        ylim=c(0,60000))
```
RENTHOME. 
```{r}
table(data_clean$RENTHOM1, useNA = "ifany")

#Condensing down to two categories. 
data_clean$RENTHOM1[data_clean$RENTHOM1==2 | data_clean$RENTHOM1==3]<-2

table(data_clean$RENTHOM1, useNA = "ifany")
mfv(data_clean$RENTHOM1)
barplot(table(data_clean$RENTHOM1, 
              useNA = "ifany"),
       col="darksalmon",
        names.arg= c("Own", "Does not\nown", "Missing"),
        ylab = "Frequency",
        ylim=c(0,120000))
```
VETERAN. 
```{r}
table(data_clean$VETERAN3, useNA = "ifany")
mfv(data_clean$VETERAN3)

barplot(table(data_clean$VETERAN3, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,120000))
```
GENHLTH. 
```{r}
table(data_clean$X_RFHLTH)
mfv(data_clean$X_RFHLTH)

barplot(table(data_clean$X_RFHLTH, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Good or better", "Fair or worse"),
        ylab = "Frequency",
        ylim=c(0,120000))
```
HLTHPLN1. 
```{r}
table(data_clean$HLTHPLN1, useNA = "ifany")
mfv(data_clean$HLTHPLN1)

barplot(table(data_clean$HLTHPLN1, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,140000))
```
MEDCOST. 
```{r}
table(data_clean$MEDCOST, useNA = "ifany")
mfv(data_clean$MEDCOST)

barplot(table(data_clean$MEDCOST, 
              useNA = "ifany"),
       col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
       ylab = "Frequency",
        ylim=c(0,150000))
```
CHECKUP. 
```{r}
table(data_clean$CHECKUP1, useNA = "ifany")

#Condensing down to "Within a year" and "more than 1 year".
data_clean$CHECKUP1[data_clean$CHECKUP1==0 | data_clean$CHECKUP1==2 | data_clean$CHECKUP1==3 | data_clean$CHECKUP1==4]<-2

#Descriptives. 
table(data_clean$CHECKUP1, useNA = "ifany")
mfv(data_clean$CHECKUP1)
barplot(table(data_clean$CHECKUP1, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,120000))
```
FLUSHOT6. 
```{r}
table(data_clean$FLUSHOT6, useNA = "ifany")
mfv(data_clean$FLUSHOT6)
barplot(table(data_clean$FLUSHOT6, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,80000))
```

X_RFHYPE5 (BLOOD PRESSURE).
```{r}
table(data_clean$X_RFHYPE5, useNA = "ifany")
mfv(data_clean$X_RFHYPE5)

barplot(table(data_clean$X_RFHYPE5, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("No", "Yes", "Missing"),
        ylab = "Frequency",
        ylim=c(0,80000))
```
X_RFCHOL (CHOLESTEROL).
```{r}
table(data_clean$X_RFCHOL, useNA = "ifany")
mfv(data_clean$X_RFCHOL)

barplot(table(data_clean$X_RFCHOL, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("No", "Yes", "Missing"),
        ylab = "Frequency",
        ylim=c(0,80000))
```
X_MICHD (HEART DISEASE).
```{r}
table(data_clean$X_MICHD, useNA = "ifany")
mfv(data_clean$X_MICHD)

barplot(table(data_clean$X_MICHD, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,120000))
```

CVDSTRK3. 
```{r}
table(data_clean$CVDSTRK3, useNA = "ifany")
mfv(data_clean$CVDSTRK3)

barplot(table(data_clean$CVDSTRK3, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,140000))
```
X_ASTHMS1. 
```{r}
table(data_clean$X_ASTHMS1, useNA = "ifany")


#Reducing down to two categories. 
data_clean$X_ASTHMS1[data_clean$X_ASTHMS1==2]<-1
data_clean$X_ASTHMS1[data_clean$X_ASTHMS1==3]<-2

#Descriptives.  
table(data_clean$X_ASTHMS1, useNA = "ifany")
mfv(data_clean$X_ASTHMS1)
barplot(table(data_clean$X_ASTHMS1, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,120000))
```

CNCER. 
```{r}
table(data_clean$CNCER, useNA = "ifany")
mfv(data_clean$CNCER)
barplot(table(data_clean$CNCER, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,80000))
```
CHCCOPD1
```{r}
table(data_clean$CHCCOPD1, useNA = "ifany")
mfv(data_clean$CHCCOPD1)

barplot(table(data_clean$CHCCOPD1, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,120000))
```
X_DRDXAR1. 
```{r}
table(data_clean$X_DRDXAR1, useNA = "ifany")
mfv(data_clean$X_DRDXAR1)

barplot(table(data_clean$X_DRDXAR1, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,80000))
```
ADDEPEV2. 
```{r}
table(data_clean$ADDEPEV2, useNA = "ifany")
mfv(data_clean$ADDEPEV2)

barplot(table(data_clean$ADDEPEV2, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,120000))
```
CHCKIDNY.
```{r}
table(data_clean$CHCKIDNY, useNA = "ifany")
mfv(data_clean$CHCKIDNY)

barplot(table(data_clean$CHCKIDNY, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,140000))
```
DIABETE3. 
```{r}
table(data_clean$DIABETE3, useNA = "ifany")

#Condensing down to "yes" and "no" responses. 
data_clean$DIABETE3[data_clean$DIABETE3==1 | data_clean$DIABETE3==2]<-1
data_clean$DIABETE3[data_clean$DIABETE3==3|data_clean$DIABETE3==4]<-2

table(data_clean$DIABETE3, useNA = "ifany")
mfv(data_clean$DIABETE3)
barplot(table(data_clean$DIABETE3, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,120000))
```
BLIND. 
```{r}
table(data_clean$BLIND, useNA = "ifany")
mfv(data_clean$BLIND)

barplot(table(data_clean$BLIND, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,140000))
```
QLACTLM2 (Activity limitations). 
```{r}
table(data_clean$QLACTLM2, useNA = "ifany")
mfv(data_clean$QLACTLM2)

barplot(table(data_clean$QLACTLM2, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,120000))
```
USEEQUIP. 
```{r}
table(data_clean$USEEQUIP, useNA = "ifany")
mfv(data_clean$USEEQUIP)

barplot(table(data_clean$USEEQUIP, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,120000))
```
DECIDE. 
```{r}
table(data_clean$DECIDE, useNA = "ifany")
mfv(data_clean$DECIDE)

barplot(table(data_clean$DECIDE, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg= c("Yes", "No", "Missing"),
        ylab = "Frequency",
        ylim=c(0,140000))
```

X_SMOKER3. 
```{r}
table(data_clean$X_SMOKER3, useNA = "ifany")

#Reducing the categories. 
data_clean$X_SMOKER3[data_clean$X_SMOKER3==1 | data_clean$X_SMOKER3==2]<-1
data_clean$X_SMOKER3[data_clean$X_SMOKER3==3]<-2
data_clean$X_SMOKER3[data_clean$X_SMOKER3==4]<-3

#Descriptives. 
table(data_clean$X_SMOKER3, useNA = "ifany")
mfv(data_clean$X_SMOKER3)
barplot(table(data_clean$X_SMOKER3, 
              useNA = "ifany"),
        col="darksalmon",
        names.arg=c("Current", "Former", "Never", "Missing"),
        ylab = "Frequency",
        ylim=c(0,80000))
```
EXRACT11. 
```{r}
#Only showing the top ten results. 
extract11 <- head(sort(table(data_clean$EXRACT11, useNA="ifany"),decreasing=TRUE),
     n=10)
print(extract11)

mfv(data_clean$EXRACT11)
barplot(extract11,
        las=2,
        names.arg=c("Walking", "No\nResponse", "Gardening", "Other", "Bicycling\nMachine", "Aerobics\nClass", "Yard Work", "Bicycling", "Golf", "Weight lifting"),
        col="darksalmon",
        cex.names=0.8)
```
```{r}
write.csv(data_clean,
          "C:/Users/mmkea/Documents/brfss_2015_data_clean.csv", 
          row.names = FALSE)
```