#Inputting Data



library(foreign)

outliers <- read.dta(file.choose())



# Creating Data



# Generate 100 observations of a set of partially related variables drawn from a normal distribution



var1 <- 2*rnorm(100)



var2 <- var1 + 2*rnorm(100)



var3 <- var1 + var2 + 2*rnorm(100)



# Let's also generate a related dichotomous variable



var4 <- var1 + var2 + var3 + 4*rnorm(100)



dummy <- ifelse(var4>=0, 1, 0)



# Now let's see what we have



objects()



# Lets get rid of one object



rm(var4)



# These are useful summary commands



stem(var1)



summary(var1)



table(dummy)



summary(outliers)



# Graphing



attach(outliers)



hist(y)



hist(y, prob=T)



rug(y)



lines(density(y, bw=100))



lines(density(y, bw=50))



# Color Graphing



hist(y, col =c( "red", "orange", "yellow"))



# Plotting, (First command plots with a defined marker; second command adds an additional variable to plot with a different marker; third command adds a legend)



plot(x1, x2, pch = 2)



points(x1, x3, pch = 12)



legend(25, 80, c("x2", "x3"), pch= c(12, 2) ) 





# Linear models



(?lm)

reg <- lm(y ~ x1 + x2 + x3, data = outliers)

summary(reg)





# R saves and readily implements many model attributes 



ouliers$yhat <- fitted(reg)



plot(reg)



# Generalized Linear Models



reg2 <- glm(y ~ x1 + x2 + x3, family=gaussian, data=outliers)

summary(reg2)



logit <- glm(dummy ~ var1 + var2 + var3, family=binomial)

probit <- glm(dummy ~ var1 + var2 + var3, family=binomial(link=probit))

summary(probit)

summary(logit)



# Rcmdr



library(Rcmdr)