# Importing Data
hw3 <- read.table("lungc.txt",header=TRUE)
attach (hw3)

# Or for csv
group <- read.csv("Group_Project.csv",header=TRUE)

# Or for STATA files
load "foreign" 	# which loads a package that is 
	       	# already installed on in your computer
                # you may have to download packages
		# from CRAN if they're not already installed
lab4 <- read.dta("lab4_p2.dta")
attach (lab4)

# Viewing data
lab4 [1:15,]
summary(data set name)
xtabs(variable~variable)  # makes a basic cross tabulation
ftable(variable, variable)  # shows frequency tables
table(variable, variable)

# Merging data 
merged <- merge(geo.new, alive, all=TRUE)

# Get data from STATA to SAS
alive <- read.dta("final1.dta")
alive [1:5,]
summary(alive)
write.table(alive, file = "alive.txt")

# Sorting Data (this is easier in other software packages)
x <- mydata$COD
y <- mydata$LOGO2UP
m <- mydata
order(y)
sortdat <- m[order(x),]

# Making new variables
emp <- rep(NA, length(employed))
emp[employed == "yes"] <- 1
emp[employed ==  "no"] <- 0

# Make a data frame
data.frame(data) 

# Afer making new variable must bind to data set
lab7.new <- cbind (lab7, ln.py, year0to10)

# Basic univariate tests
x <- matrix (c (2, 4,
		13, 15),2)
chisq.test(x, correct = FALSE)

OR = apply(z, 1,
	function(x) x[1] * (n2 - x[2]) / (x[2] * (n1 - x[1]))
	)
lnOR <- log(OR)

dat <- read.csv("task1.csv", header = FALSE)
dat2 <- dat[,1 : 10]
dat2 <- dat2[complete.cases(dat2),]
vec1 <- as.vector(unlist(dat2))

chisq.test(table (vec1), correct = FALSE)

fisher.test(dat, alternative = "less")

wilcox.test(x,y, alternative = c("two.sided"))

# Create spline term:
CODspl <- (COD-5000)*(COD>4999)

# Plotting lines
fitcii <- lm(formula = LOGO2~CODc+CODc2)$fit
plot(LOGO2~COD,cex=0.6,main="1cii: Quadratic regression")
lines(fitcii~COD,col="blue")
abline(h=0)

# Making Dummy Variables
c.mos <- factor(tot_mos)
c.mos
cbind (tot_mos, model.matrix(~c.mos-1))

# Or use factor(tot_mos) in model

# Subsetting Data 
men <- subset (lab7, ma0fe1 == 0)
men
women <- subset (lab7, ma0fe1 == 1)
women

# Changing reference level with catagorical data
new <- relevel( factor(tot_mos), ref = "12") 

# Lowess Smoother
       # install sma
plot(jitter(nMedExp,0.1), jitter(case,0.1))
plot.smooth.line (jitter(nMedExp,0.1), jitter(case,0.1))

# ANOVA Table
fit.small <- (lm (disease. ~ crowding. + airqual.))
fit.big <- (lm (disease. ~ crowding. + airqual. + education.))
anova(fit.small)
anova(fit.big)
anova(fit.small, fit.big, test = "F")

# Linear models
summary (lm (cd4~bl.age + other + variables))

# Cox
load Survival
model <- (coxph (Surv(fu.time, sc) ~ ma0fe1 + sexyear))

# Logistic Regression
load "Design"
lrm(formula, data, subset, na.action=na.delete, method="lrm.fit",
    model=FALSE, x=FALSE, y=FALSE, linear.predictors=TRUE, se.fit=FALSE, 
    penalty=0, penalty.matrix, tol=1e-7, 
    strata.penalty=0, var.penalty=c('simple','sandwich'),
    weights, normwt, ...)
      #or
mod <- glm (I(n.sym/n.mice)~dose, family = binomial (link = "logit"))

# Poisson
pfit <- glm(cases~factor(cals)+factor(train)+factor(bank), 
	family=poisson (link=log), data=p1, offset = log(mm))

# Mixed Models
load "lme"  #or
load "nlme"   #for non linear mixed models


# Big thing of graphs
pairs (~disease. + crowding. + airqual. + education. + nutrition. + smoking)

# Residuals
resid <- model1$resid
plot (salary ~ service, ylab = "Salary", xlab = "Service")
plot (resid ~ fit, xlab = "Fitted Values", ylab = "Residuals")
abline (h = 0)
plot (resid ~ service, ylab = "Residuals", xlab = "Service")
abline (h = 0)

# Influence
hii <- lm.influence(model1)$hat
stud <- 1 – hii
cook <- cooks.distance(model1)
stem (hii, 2.5)
sresid <- ls.diag(model1)$stud.res
plot (model1, which = 5)   # makes funky graph
round (cbind (hw4, hii, stud, cook, sresid), 3)

# VIF
library(Design)
vif(big)